Dissertations / Theses on the topic 'Online learning algorithms'
To see the other types of publications on this topic, follow the link: Online learning algorithms.
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Online learning algorithms.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
1
Harrington, Edward Francis. "Aspects of online learning /." View thesis entry in Australian Digital Theses Program, 2004. http://thesis.anu.edu.au/public/adt-ANU20060328.160810/index.html.
Full text
2
Harrington, Edward, and edwardharrington@homemail com au. "Aspects of Online Learning." The Australian National University. Research School of Information Sciences and Engineering, 2004. http://thesis.anu.edu.au./public/adt-ANU20060328.160810.
Full textAbstract:
Online learning algorithms have several key advantages compared to their batch learning algorithm counterparts: they are generally more memory efficient, and computationally mor efficient; they are simpler to implement; and they are able to adapt to changes where the learning model is time varying. Online algorithms because of their simplicity are very appealing to practitioners. his thesis investigates several online learning algorithms and their application. The thesis has an underlying theme of the idea of combining several simple algorithms to give better performance. In this thesis we investigate: combining weights, combining hypothesis, and (sort of) hierarchical combining.¶
Firstly, we propose a new online variant of the Bayes point machine (BPM), called the online Bayes point machine (OBPM). We study the theoretical and empirical performance of the OBPm algorithm. We show that the empirical performance of the OBPM algorithm is comparable with other large margin classifier methods such as the approximately large margin algorithm (ALMA) and methods which maximise the margin explicitly, like the support vector machine (SVM). The OBPM algorithm when used with a parallel architecture offers potential computational savings compared to ALMA. We compare the test error performance of the OBPM algorithm with other online algorithms: the Perceptron, the voted-Perceptron, and Bagging. We demonstrate that the combinationof the voted-Perceptron algorithm and the OBPM algorithm, called voted-OBPM algorithm has better test error performance than the voted-Perceptron and Bagging algorithms. We investigate the use of various online voting methods against the problem of ranking, and the problem of collaborative filtering of instances. We look at the application of online Bagging and OBPM algorithms to the telecommunications problem of channel equalization. We show that both online methods were successful at reducing the effect on the test error of label flipping and additive noise.¶
Secondly, we introduce a new mixture of experts algorithm, the fixed-share hierarchy (FSH) algorithm. The FSH algorithm is able to track the mixture of experts when the switching rate between the best experts may not be constant. We study the theoretical aspects of the FSH and the practical application of it to adaptive equalization. Using simulations we show that the FSH algorithm is able to track the best expert, or mixture of experts, in both the case where the switching rate is constant and the case where the switching rate is time varying.
3
Pasteris, S. U. "Efficient algorithms for online learning over graphs." Thesis, University College London (University of London), 2016. http://discovery.ucl.ac.uk/1516210/.
Full textAbstract:
In this thesis we consider the problem of online learning with labelled graphs, in particular designing algorithms that can perform this problem quickly and with low memory requirements. We consider the tasks of Classification (in which we are asked to predict the labels of vertices) and Similarity Prediction (in which we are asked to predict whether two given vertices have the same label). The first half of the thesis considers non- probabilistic online learning, where there is no probability distribution on the labelling and we bound the number of mistakes of an algorithm by a function of the labelling's complexity (i.e. its "naturalness"), often the cut- size. The second half of the thesis considers probabilistic machine learning in which we have a known probability distribution on the labelling. Before considering probabilistic online learning we first analyse the junction tree algorithm, on which we base our online algorithms, and design a new ver- sion of it, superior to the otherwise current state of the art. Explicitly, the novel contributions of this thesis are as follows: • A new algorithm for online prediction of the labelling of a graph which has better performance than previous algorithms on certain graph and labelling families. • Two algorithms for online similarity prediction on a graph (a novel problem solved in this thesis). One performs very well whilst the other not so well but which runs exponentially faster. • A new (better than before, in terms of time and space complexity) state of the art junction tree algorithm, as well as an application of it to the problem of online learning in an Ising model. • An algorithm that, in linear time, finds the optimal junction tree for online inference in tree-structured Ising models, the resulting online junction tree algorithm being far superior to the previous state of the art. All claims in this thesis are supported by mathematical proofs.
4
Packer, Heather S. "Evolving ontologies with online learning and forgetting algorithms." Thesis, University of Southampton, 2011. https://eprints.soton.ac.uk/194923/.
Full textAbstract:
Agents that require vocabularies to complete tasks can be limited by static vocabularies which cannot evolve to meet unforeseen domain tasks, or reflect its changing needs or environment. However, agents can benefit from using evolution algorithms to evolve their vocabularies, namely the ability to support new domain tasks. While an agent can capitalise on being able support more domain tasks, using existing techniques can hinder them because they do not consider the associated costs involved with evolving an agent's ontology. With this motivation, we explore the area of ontology evolution in agent systems, and focus on the reduction of the costs associated with an evolving ontology. In more detail, we consider how an agent can reduce the costs of evolving an ontology, these include costs associated with: the acquisition of new concepts; processing new concepts; the increased memory usage from storing new concepts; and the removal of unnecessary concepts. Previous work reported in the literature has largely failed to analyse these costs in the context of evolving an agent's ontology. Against this background, we investigate and develop algorithms to enable agents to evolve their ontologies. More specifically, we present three online evolution algorithms that enable agents to: i) augment domain related concepts, ii) use prediction to select concepts to learn, and iii) prune unnecessary concepts from their ontology, with the aim to reduce the costs associated with the acquisition, processing and storage of acquired concepts. In order to evaluate our evolution algorithms, we developed an agent framework which enables agents to use these algorithms and measure an agent's performance. Finally, our empirical evaluation shows that our algorithms are successful in reducing the costs associated with evolving an agent's ontology.
5
Li, Le. "Online stochastic algorithms." Thesis, Angers, 2018. http://www.theses.fr/2018ANGE0031.
Full textAbstract:
Cette thèse travaille principalement sur trois sujets. Le premier concentre sur le clustering en ligne dans lequel nous présentons un nouvel algorithme stochastique adaptatif pour regrouper des ensembles de données en ligne. Cet algorithme repose sur l'approche quasi-bayésienne, avec une estimation dynamique (i.e., dépendant du temps) du nombre de clusters. Nous prouvons que cet algorithme atteint une borne de regret de l'ordre et que cette borne est asymptotiquement minimax sous la contrainte sur le nombre de clusters. Nous proposons aussi une implémentation par RJMCMC. Le deuxième sujet est lié à l'apprentissage séquentiel des courbes principales qui cherche à résumer une séquence des données par une courbe continue. Pour ce faire, nous présentons une procédure basée sur une approche maximum a posteriori pour le quasi-posteriori de Gibbs. Nous montrons que la borne de regret de cet algorithme et celui de sa version adaptative est sous-linéaire en l'horizon temporel T. En outre, nous proposons une implémentation par un algorithme glouton local qui intègre des éléments de sleeping experts et de bandit à plusieurs bras. Le troisième concerne les travaux qui visent à accomplir des tâches pratiques au sein d'iAdvize, l'entreprise qui soutient cette thèse. Il inclut l'analyse des sentiments pour les messages textuels et l'implémentation de chatbot dans lesquels la première est réalisé par les méthodes classiques dans la fouille de textes et les statistiques et la seconde repose sur le traitement du langage naturel et les réseaux de neurones artificielsThis thesis works mainly on three subjects. The first one is online clustering in which we introduce a new and adaptive stochastic algorithm to cluster online dataset. It relies on a quasi-Bayesian approach, with a dynamic (i.e., time-dependent) estimation of the (unknown and changing) number of clusters. We prove that this algorithm has a regret bound of the order of and is asymptotically minimax under the constraint on the number of clusters. A RJMCMC-flavored implementation is also proposed. The second subject is related to the sequential learning of principal curves which seeks to represent a sequence of data by a continuous polygonal curve. To this aim, we introduce a procedure based on the MAP of Gibbs-posterior that can give polygonal lines whose number of segments can be chosen automatically. We also show that our procedure is supported by regret bounds with sublinear remainder terms. In addition, a greedy local search implementation that incorporates both sleeping experts and multi-armed bandit ingredients is presented. The third one concerns about the work which aims to fulfilling practical tasks within iAdvize, the company which supports this thesis. It includes sentiment analysis for textual messages by using methods in both text mining and statistics, and implementation of chatbot based on nature language processing and neural networks
6
Minerva, Michela. "Automated Configuration of Offline/Online Algorithms: an Empirical Model Learning Approach." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/22649/.
Full textAbstract:
The energy management system is the intelligent core of a virtual power plant and it manages power flows among units in the grid. This implies dealing with optimization under uncertainty because entities such as loads and renewable energy resources have stochastic behaviors. A hybrid offline/online optimization technique can be applied in such problems to ensure efficient online computation.
This work devises an approach that integrates machine learning and optimization models to perform automatic algorithm configuration. It is inserted as the top component in a two-level hierarchical optimization system for the VPP, with the goal of configuring the low-level offline/online optimizer.
Data from the low-level algorithm is used for training machine learning models - decision trees and neural networks – that capture the highly complex behavior of both the controlled VPP and the offline/online optimizer. Then, Empirical Model Learning is adopted to build the optimization problem, integrating usual mathematical programming and ML models.
The proposed approach successfully combines optimization and machine learning in a data-driven and flexible tool that performs automatic configuration and forecasting of the low-level algorithm for unseen input instances.
7
Pesaranghader, Ali. "A Reservoir of Adaptive Algorithms for Online Learning from Evolving Data Streams." Thesis, Université d'Ottawa / University of Ottawa, 2018. http://hdl.handle.net/10393/38190.
Full textAbstract:
Continuous change and development are essential aspects of evolving environments and applications, including, but not limited to, smart cities, military, medicine, nuclear reactors, self-driving cars, aviation, and aerospace. That is, the fundamental characteristics of such environments may evolve, and so cause dangerous consequences, e.g., putting people lives at stake, if no reaction is adopted. Therefore, learning systems need to apply intelligent algorithms to monitor evolvement in their environments and update themselves effectively. Further, we may experience fluctuations regarding the performance of learning algorithms due to the nature of incoming data as it continuously evolves. That is, the current efficient learning approach may become deprecated after a change in data or environment. Hence, the question 'how to have an efficient learning algorithm over time against evolving data?' has to be addressed. In this thesis, we have made two contributions to settle the challenges described above.
In the machine learning literature, the phenomenon of (distributional) change in data is known as concept drift. Concept drift may shift decision boundaries, and cause a decline in accuracy. Learning algorithms, indeed, have to detect concept drift in evolving data streams and replace their predictive models accordingly. To address this challenge, adaptive learners have been devised which may utilize drift detection methods to locate the drift points in dynamic and changing data streams. A drift detection method able to discover the drift points quickly, with the lowest false positive and false negative rates, is preferred. False positive refers to incorrectly alarming for concept drift, and false negative refers to not alarming for concept drift. In this thesis, we introduce three algorithms, called as the Fast Hoeffding Drift Detection Method (FHDDM), the Stacking Fast Hoeffding Drift Detection Method (FHDDMS), and the McDiarmid Drift Detection Methods (MDDMs), for detecting drift points with the minimum delay, false positive, and false negative rates. FHDDM is a sliding window-based algorithm and applies Hoeffding’s inequality (Hoeffding, 1963) to detect concept drift. FHDDM slides its window over the prediction results, which are either 1 (for a correct prediction) or 0 (for a wrong prediction). Meanwhile, it compares the mean of elements inside the window with the maximum mean observed so far; subsequently, a significant difference between the two means, upper-bounded by the Hoeffding inequality, indicates the occurrence of concept drift. The FHDDMS extends the FHDDM algorithm by sliding multiple windows over its entries for a better drift detection regarding the detection delay and false negative rate. In contrast to FHDDM/S, the MDDM variants assign weights to their entries, i.e., higher weights are associated with the most recent entries in the sliding window, for faster detection of concept drift. The rationale is that recent examples reflect the ongoing situation adequately. Then, by putting higher weights on the latest entries, we may detect concept drift quickly. An MDDM algorithm bounds the difference between the weighted mean of elements in the sliding window and the maximum weighted mean seen so far, using McDiarmid’s inequality (McDiarmid, 1989). Eventually, it alarms for concept drift once a significant difference is experienced. We experimentally show that FHDDM/S and MDDMs outperform the state-of-the-art by representing promising results in terms of the adaptation and classification measures.
Due to the evolving nature of data streams, the performance of an adaptive learner, which is defined by the classification, adaptation, and resource consumption measures, may fluctuate over time. In fact, a learning algorithm, in the form of a (classifier, detector) pair, may present a significant performance before a concept drift point, but not after. We define this problem by the question 'how can we ensure that an efficient classifier-detector pair is present at any time in an evolving environment?' To answer this, we have developed the Tornado framework which runs various kinds of learning algorithms simultaneously against evolving data streams. Each algorithm incrementally and independently trains a predictive model and updates the statistics of its drift detector. Meanwhile, our framework monitors the (classifier, detector) pairs, and recommends the efficient one, concerning the classification, adaptation, and resource consumption performance, to the user. We further define the holistic CAR measure that integrates the classification, adaptation, and resource consumption measures for evaluating the performance of adaptive learning algorithms. Our experiments confirm that the most efficient algorithm may differ over time because of the developing and evolving nature of data streams.
8
Al-Janabi, Mohammed Fadhil Zamil. "Detection of suspicious URLs in online social networks using supervised machine learning algorithms." Thesis, Keele University, 2018. http://eprints.keele.ac.uk/5581/.
Full textAbstract:
This thesis proposes the use of several supervised machine learning classification models that were built to detect the distribution of malicious content in OSNs. The main focus was on ensemble learning algorithms such as Random Forest, gradient boosting trees, extra trees, and XGBoost. Features were used to identify social network posts that contain malicious URLs derived from several sources, such as domain WHOIS record, web page content, URL lexical and redirection data, and Twitter metadata. The thesis describes a systematic analysis of the hyper-parameters of tree-based models. The impact of key parameters, such as the number of trees, depth of trees and minimum size of leaf nodes on classification performance, was assessed. The results show that controlling the complexity of Random Forest classifiers applied to social media spam is essential to avoid overfitting and optimise performance. The model complexity could be reduced by removing uninformative features, as the complexity they add to the model is greater than the advantages they give to the model to make decisions. Moreover, model-combining methods were tested, which are the voting and stacking methods. Both show advantages and disadvantages; however, in general, they appear to provide a statistically significant improvement in comparison to the highest singular model. The critical benefit of applying the stacking method to automate the model selection process is that it is effective in giving more weight to more topperforming models and less affected by weak ones. Finally, 'SuspectRate', an online malicious URL detection system, was built to offer a service to give a suspicious probability of tweets with attached URLs. A key feature of this system is that it can dynamically retrain and expand current models.
9
Zheng, Zhilin. "Learning Group Composition and Re-composition in Large-scale Online Learning Contexts." Doctoral thesis, Humboldt-Universität zu Berlin, 2017. http://dx.doi.org/10.18452/18412.
Full textAbstract:
Die Erforschung der Zusammenstellung kleiner Lerngruppen beschäftigt sich mit dem Problem, eine passende Gruppenzusammensetzung in einer Population von Lernern zu finden, die jeder Gruppe optimalen Nutzen bringen könnte. In letzter Zeit sind viele Studien zu diesem Problem der Kleingruppenzusammenstellung durchgeführt worden. Allerdings waren diese Forschungen nur selten auf den Kontext großer Lerner-Populationen ausgerichtet. Angesichts des zunehmenden Aufkommens von MOOCs muss jedoch das Problem der Gruppenzusammenstellung entsprechend erweitert betrachtet werden, und zwar mit neuen Forschungen, die den Kontext derartig großer Lerner-Populationen berücksichtigen. Anders als in Klassenzimmer-Settings könnte die beobachtete hohe Abbruchquote in MOOCs in einer Unterbesetzung der Gruppengröße resultieren und könnte somit viele Lerner dazu bringen, neue Gruppen zu bilden. Zusätzlich zur Gruppenzusammenstellung muss daher die Gruppenneuzusammenstellung als neues Thema in aktuellen Kontexten großer Lerner-Populationen ebenfalls erforscht werden.
Die Untersuchungen der vorliegenden Arbeit gliedern sich in zwei Teile. Der erste Teil beschäftigt sich mit Gruppenzusammenstellung. In diesem Teil stelle ich einen diskreten-PSO Algorithmus zur Zusammenstellung kleiner Lerngruppen vor und vergleiche bislang bestehende Gruppenzusammenstellungs-Algorithmen unter den Gesichtspunkten Zeitaufwand und Gruppierungsqualität. Um Gruppenzusammenstellung in MOOCs anzuwenden wurde ein Gruppenzusammenstellungsexperiment in einem MOOC durchgeführt. Die Hauptergebnisse deuten darauf hin, dass die Gruppenzusammenstellung die Abbruchsquote reduzieren kann, jedoch lediglich einen sehr schwachen Bezug zur Lernperformanz der Lerner aufweist. Der zweite Teil beschäftigt sich mit Gruppenneuzusammenstellung. Die vorliegende Arbeit stellt eine datengesteuerte Herangehensweise vor, die umfassenden Gebrauch von Gruppeninteraktionsdaten macht sowie Gruppendynamik mit einbezieht. Mittels einer in einem Simulationsexperiment durchgeführten Evaluation zeigen sich die Vorteile dieses Verfahrens: Der Lerngruppenzusammenhalt wird verbessert und die Abbruchsquote im Vergleich zu einer Zufallsverteilung reduziert. Darüberhinaus wurde hier ein Gruppen-Lern-Werkzeug entwickelt und für die Praxis vorbereitet, das die Anforderungen des geforderten Ansatzes der Gruppenneuzusammenstellung erfüllt.Small learning group composition addresses the problem of seeking such matching among a population of students that it could bring each group optimal benefits. Recently, many studies have been conducted to address this small group composition problem. Nevertheless, the focus of such a body of research has rarely been cast to large-scale contexts. Due to the recent come of MOOCs, the topic of group composition needs to be accordingly extended with new investigations in such large learning contexts. Different from classroom settings, the reported high drop-out rate of MOOCs could result in group’s incompletion in size and thus might compel many students to compose new groups. Thus, in addition to group composition, group re-composition as a new topic needs to be studied in current large-scale learning contexts as well. In this thesis, the research is structured in two stages. The first stage is group composition. In this part, I proposed a discrete-PSO algorithm to compose small learning groups and compared the existing group composition algorithms from the perspectives of time cost and grouping quality. To implement group composition in MOOCs, a group composition experiment was conducted in a MOOC. The main results indicate that group composition can reduce drop-out rate, yet has a very weak association with students’ learning performance. The second stage is to cope with group re-composition. This thesis suggests a data-driven approach that makes full use of group interaction data and accounts for group dynamics. Through evaluation in a simulation experiment, it shows its advantages of bringing us more cohesive learning groups and reducing the drop-out rate compared to a random condition. Apart from these, a group learning tool that fulfills the goals of the proposed group re-composition approach has been developed and is made ready for practice.
10
Heidari, Fariba. "Quality of service routing using decentralized learning." Thesis, McGill University, 2009. http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=115672.
Full textAbstract:
This thesis presents several decentralized, learning-based algorithms for on-line routing of bandwidth guaranteed paths. The presented routing algorithms do not need any a-priori knowledge of traffic demand; they use only their locally observed events and update their routing policy using learning schemes. The employed learning algorithms are either learning automata or the multi-armed bandit algorithms. We investigate the asymptotic behavior of the proposed routing algorithms and prove the convergence of one of them to the user equilibrium. Discrete event simulation results show the merit of these algorithms in terms of improving the resource utilization and increasing the network admissibility compared with shortest path routing.We investigate the performance degradation due to decentralized routing as opposed to centralized optimal routing policies in practical scenarios. The system optimal and the Nash bargaining solutions are two centralized benchmarks used in this study. We provide nonlinear programming formulations of these problems along with a distributed recursive approach to compute the solutions. An on-line partially-decentralized control architecture is also proposed to achieve the system optimal and the Nash bargaining solution performances. Numerical results in some practical scenarios with well engineered networks, where the network resources and traffic demand are well matched, indicate that decentralized learning techniques provide efficient, stable and scalable approaches for routing the bandwidth guaranteed paths.
In the context of on-line learning, we propose a new algorithm to track the best action-selection policy when it abruptly changes over time. The proposed algorithm employs change detection mechanisms to detect the sudden changes and restarts the learning process on the detection of an abrupt change. The performance analysis of this study reveals that when all the changes are detectable by the change detection mechanism, the proposed tracking the best action-selection policy algorithm is rate optimal. On-line routing of bandwidth guaranteed paths with the potential occurrence of network shocks such as significant changes in the traffic demand is one of the applications of the devised algorithm. Simulation results show the merit of the proposed algorithm in tracking the optimal routing policy when it abruptly changes.
11
Melki, Gabriella A. "Fast Online Training of L1 Support Vector Machines." VCU Scholars Compass, 2016. http://scholarscompass.vcu.edu/etd/4282.
Full textAbstract:
This thesis proposes a novel experimental environment (non-linear stochastic gradient descent, NL-SGD), as well as a novel online learning algorithm (OL SVM), for solving a classic nonlinear Soft Margin L1 Support Vector Machine (SVM) problem using a Stochastic Gradient Descent (SGD) algorithm. The NL-SGD implementation has a unique method of random sampling and alpha calculations. The developed code produces a competitive accuracy and speed in comparison with the solutions of the Direct L2 SVM obtained by software for Minimal Norm SVM (MN-SVM) and Non-Negative Iterative Single Data Algorithm (NN-ISDA). The latter two algorithms have shown excellent performances on large datasets; which is why we chose to compare NL-SGD and OL SVM to them. All experiments have been done under strict double (nested) cross-validation, and the results are reported in terms of accuracy and CPU times. OL SVM has been implemented within MATLAB and is compared to the classic Sequential Minimal Optimization (SMO) algorithm implemented within MATLAB's solver, fitcsvm. The experiments with OL SVM have been done using k-fold cross-validation and the results reported in % error and % speedup of CPU Time.
12
Holmgren, Faghihi Josef, and Paul Gorgis. "Time efficiency and mistake rates for online learning algorithms : A comparison between Online Gradient Descent and Second Order Perceptron algorithm and their performance on two different data sets." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-260087.
Full textAbstract:
This dissertation investigates the differences between two different online learning algorithms: Online Gradient Descent (OGD) and Second-Order Perceptron (SOP) algorithm, and how well they perform on different data sets in terms of mistake rate, time cost and number of updates. By studying different online learning algorithms and how they perform in different environments will help understand and develop new strategies to handle further online learning tasks. The study includes two different data sets, Pima Indians Diabetes and Mushroom, together with the LIBOL library for testing. The results in this dissertation show that Online Gradient Descent performs overall better concerning the tested data sets. In the first data set, Online Gradient Descent recorded a notably lower mistake rate. For the second data set, although it recorded a slightly higher mistake rate, the algorithm was remarkably more time efficient compared to Second-Order Perceptron. Future work would include a wider range of testing with more, and different, data sets as well as other relative algorithms. This will lead to better result and higher credibility.Den här avhandlingen undersöker skillnaden mellan två olika “online learning”-algoritmer: Online Gradient Descent och Second-Order Perceptron, och hur de presterar på olika datasets med fokus på andelen felklassificeringar, tidseffektivitet och antalet uppdateringar. Genom att studera olika “online learning”-algoritmer och hur de fungerar i olika miljöer, kommer det hjälpa till att förstå och utveckla nya strategier för att hantera vidare “online learning”-problem. Studien inkluderar två olika dataset, Pima Indians Diabetes och Mushroom, och använder biblioteket LIBOL för testning. Resultatet i denna avhandling visar att Online Gradient Descent presterar bättre överlag på de testade dataseten. För det första datasetet visade Online Gradient Descent ett betydligt lägre andel felklassificeringar. För det andra datasetet visade OGD lite högre andel felklassificeringar, men samtidigt var algoritmen anmärkningsvärt mer tidseffektiv i jämförelse med Second-Order Perceptron. Framtida studier inkluderar en bredare testning med mer, och olika, datasets och andra relaterade algoritmer. Det leder till bättre resultat och höjer trovärdigheten.
13
Dennis, Aaron W. "Algorithms for Learning the Structure of Monotone and Nonmonotone Sum-Product Networks." BYU ScholarsArchive, 2016. https://scholarsarchive.byu.edu/etd/6188.
Full textAbstract:
The sum-product network (SPN) is a recently-proposed generative, probabilistic model that is guaranteed to compute any joint or any marginal probability in time linear in the size of the model. An SPN is represented as a directed, acyclic graph (DAG) of sum and product nodes, with univariate probability distributions at the leaves. It is important to learn the structure of this DAG since the set of distributions representable by an SPN is constrained by it. We present the first batch structure learning algorithm for SPNs and show its advantage over learning the parameters of an SPN with fixed architecture. We propose a search algorithm for learning the structure of an SPN and show that its ability to learn a DAG-structured SPN makes it better for some tasks than algorithms that only learn tree-structured SPNs. We adapt the structure search algorithm to learn the structure of an SPN in the online setting and show that two other methods for online SPN structure learning are slower or learn models with lower likelihood. We also propose to combine SPNs with an autoencoder to model image data; this application of SPN structure learning shows that both models benefit from being combined.We are also the first to propose a distinction between nonmonotone and monotone SPNs, or SPNs with negative edge-weights and those without, respectively. We prove several important properties of nonmonotone SPNs, propose algorithms for learning a special class of nonmonotone SPN called the twin SPNs, and show that allowing negative edge-weights can help twin SPNs model some distributions more compactly than monotone SPNs.
14
Crocomo, Márcio Kassouf. "Um algoritmo evolutivo para aprendizado on-line em jogos eletrônicos." Universidade de São Paulo, 2008. http://www.teses.usp.br/teses/disponiveis/55/55134/tde-09052008-160236/.
Full textAbstract:
Este trabalho verifica a possibilidade de se aplicar Algoritmos Evolutivos no aprendizado on-line de jogos. Alguns autores concordam que Algoritmos Evolutivos não são aplicáveis na prática para se atingir o objetivo em questão. É com a intenção de contestar a veracidade desta afirmação que foi desenvolvido o presente trabalho. Para atingir o objetivo proposto, foi desenvolvido um jogo de computador, no qual o algoritmo de aprendizado gera estratégias inteligentes e adaptativas para os caracteres não controlados pelo jogador através de um algoritmo evolutivo. Desta forma, a função do algoritmo evolutivo é fazer com que a estratégia utilizada pelo computador se adapte à estratégia utilizada pelo usuário a cada vez que joga. É apresentada uma revisão bibliográfica a respeito de Computação Evolutiva e as técnicas utilizadas para implementar comportamentos inteligentes para os caracteres controlados por computador nos jogos atuais, esclarecendo suas vantagens, desvantagens e algumas possíveis aplicações. São também explicados o jogo e os algoritmos implementados, assim como os experimentos realizados e seus resultados. Por fim, é feita uma comparação do algoritmo evolutivo final com uma outra técnica de adaptação, chamada Dynamic Scripting. Assim, este trabalho oferece contribuições para o campo de Computação Evolutiva e Inteligência Artificial aplicada a jogosThe goal of this work is to verify if it is possible to apply Evolutionary Algorithms to online learning in computer games. Some authors agree that evolutionary algorithms do not work properly in that case. With the objective of contesting this affirmation, this work was performed. To accomplish the goal of this work, a computer game was developed, in which the learning algorithm must create intelligent and adaptive strategies to control the non-player characters using an evolutionary algorithm. Therefore, the aim of the evolutionary algorithm is to adapt the strategy used by the computer according to the player\'s actions during the game. A review on Evolutionary Computation and the techniques used to produce intelligent behaviors for the computer controlled characters in modern game is presented, exposing the advantages, the problems and some applications of each technique. The proposed game is also explained, together with the implemented algorithms, the experiments and the obtained results. Finally, it is presented a comparison between the implemented algorithm and the Dynamic Script technique. Thus, this work offers contributions to the fields of Evolutionary Computation and Artificial Intelligence applied to games
15
Al, Rawashdeh Khaled. "Toward a Hardware-assisted Online Intrusion Detection System Based on Deep Learning Algorithms for Resource-Limited Embedded Systems." University of Cincinnati / OhioLINK, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1535464571843315.
Full text
16
Rekanar, Kaavya. "Text Classification of Legitimate and Rogue online Privacy Policies : Manual Analysis and a Machine Learning Experimental Approach." Thesis, Blekinge Tekniska Högskola, Institutionen för datalogi och datorsystemteknik, 2016. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-13363.
Full text
17
Frery, Jordan. "Ensemble Learning for Extremely Imbalced Data Flows." Thesis, Lyon, 2019. http://www.theses.fr/2019LYSES034.
Full textAbstract:
L'apprentissage machine est l'étude de la conception d'algorithmes qui apprennent à partir des données d'apprentissage pour réaliser une tâche spécifique. Le modèle résultant est ensuite utilisé pour prédire de nouveaux points de données (invisibles) sans aucune aide extérieure. Ces données peuvent prendre de nombreuses formes telles que des images (matrice de pixels), des signaux (sons,...), des transactions (âge, montant, commerçant,...), des journaux (temps, alertes, ...). Les ensembles de données peuvent être définis pour traiter une tâche spécifique telle que la reconnaissance d'objets, l'identification vocale, la détection d'anomalies, etc. Dans ces tâches, la connaissance des résultats escomptés encourage une approche d'apprentissage supervisé où chaque donnée observée est assignée à une étiquette qui définit ce que devraient être les prédictions du modèle. Par exemple, dans la reconnaissance d'objets, une image pourrait être associée à l'étiquette "voiture" qui suggère que l'algorithme d'apprentissage doit apprendre qu'une voiture est contenue dans cette image, quelque part. Cela contraste avec l'apprentissage non supervisé où la tâche à accomplir n'a pas d'étiquettes explicites. Par exemple, un sujet populaire dans l'apprentissage non supervisé est de découvrir les structures sous-jacentes contenues dans les données visuelles (images) telles que les formes géométriques des objets, les lignes, la profondeur, avant d'apprendre une tâche spécifique. Ce type d'apprentissage est évidemment beaucoup plus difficile car il peut y avoir un nombre infini de concepts à saisir dans les données. Dans cette thèse, nous nous concentrons sur un scénario spécifique du cadre d'apprentissage supervisé : 1) l'étiquette d'intérêt est sous-représentée (p. ex. anomalies) et 2) l'ensemble de données augmente avec le temps à mesure que nous recevons des données d'événements réels (p. ex. transactions par carte de crédit). En fait, ces deux problèmes sont très fréquents dans le domaine industriel dans lequel cette thèse se dérouleMachine learning is the study of designing algorithms that learn from trainingdata to achieve a specific task. The resulting model is then used to predict overnew (unseen) data points without any outside help. This data can be of manyforms such as images (matrix of pixels), signals (sounds,...), transactions (age,amount, merchant,...), logs (time, alerts, ...). Datasets may be defined to addressa specific task such as object recognition, voice identification, anomaly detection,etc. In these tasks, the knowledge of the expected outputs encourages a supervisedlearning approach where every single observed data is assigned to a label thatdefines what the model predictions should be. For example, in object recognition,an image could be associated with the label "car" which suggests that the learningalgorithm has to learn that a car is contained in this picture, somewhere. This is incontrast with unsupervised learning where the task at hand does not have explicitlabels. For example, one popular topic in unsupervised learning is to discoverunderlying structures contained in visual data (images) such as geometric formsof objects, lines, depth, before learning a specific task. This kind of learning isobviously much harder as there might be potentially an infinite number of conceptsto grasp in the data. In this thesis, we focus on a specific scenario of thesupervised learning setting: 1) the label of interest is under represented (e.g.anomalies) and 2) the dataset increases with time as we receive data from real-lifeevents (e.g. credit card transactions). In fact, these settings are very common inthe industrial domain in which this thesis takes place
18
Berthold, Oswald. "Robotic self-exploration and acquisition of sensorimotor skills." Doctoral thesis, Humboldt-Universität zu Berlin, 2020. http://dx.doi.org/10.18452/21480.
Full textAbstract:
Die Interaktion zwischen Maschinen und ihrer Umgebung sollte
zuverlässig, sicher und ökologisch adequat sein. Um das in komplexen
Szenarien langfristig zu gewährleisten, wird eine Theorie adaptiven
Verhaltens benötigt. In der Entwicklungsrobotik und verkörperten
künstlichen Intelligenz wird Verhalten als emergentes Phänomen auf der
fortlaufenden dynamischen Interaktion zwischen Agent, Körper und
Umgebung betrachtet.
Die Arbeit untersucht Roboter, die in der Lage sind, schnell und
selbständig einfache Bewegungen auf Grundlage sensomotorischer
Information zu erlernen. Das langfristige Ziel dabei ist die
Wiederverwendung gelernter Fertigkeiten in späteren Lernprozessen um
damit ein komplexes Interaktionsrepertoire mit der Welt entstehen zu
lassen, das durch Entwicklungsprozesse vollständig und fortwährend
adaptiv in der sensomotorischen Erfahrung verankert ist.
Unter Verwendung von Methoden des maschinellen Lernens, der
Neurowissenschaft, Statistik und Physik wird die Frage in die
Komponenten Repräsentation, Exploration, und Lernen zerlegt. Es wird
ein Gefüge für die systematische Variation und Evaluation von Modellen
errichtet. Das vorgeschlagene Rahmenwerk behandelt die prozedurale
Erzeugung von Hypothesen als Flussgraphen über einer festen Menge von
Funktionsbausteinen, was die Modellsuche durch nahtlose Anbindung über
simulierte und physikalische Systeme hinweg ermöglicht.
Ein Schwerpunkt der Arbeit liegt auf dem kausalen Fussabdruck des
Agenten in der sensomotorischen Zeit. Dahingehend wird ein
probabilistisches graphisches Modell vorgeschlagen um Infor-
mationsflussnetzwerke in sensomotorischen Daten zu repräsentieren. Das
Modell wird durch einen auf informationtheoretischen Grössen
basierenden Lernalgorithmus ergänzt. Es wird ein allgemeines Modell
für Entwicklungslernen auf Basis von Echtzeit-Vorhersagelernen
präsentiert und anhand dreier Variationen näher besprochen.The interaction of machines with their environment should be reliable, safe, and ecologically adequate. To ensure this over long-term complex scenarios, a theory of adaptive behavior is needed. In developmental robotics, and embodied artificial intelligence behavior is regarded as a phenomenon that emerges from an ongoing dynamic interaction between entities called agent, body, and environment. The thesis investigates robots that are able to learn rapidly and on their own, how to do primitive motions, using sensorimotor information. The long-term goal is to reuse acquired skills when learning other motions in the future, and thereby grow a complex repertoire of possible interactions with the world, that is fully grounded in, and continually adapted to sensorimotor experience through developmental processes. Using methods from machine learning, neuroscience, statistics, and physics, the question is decomposed into the relationship of representation, exploration, and learning. A framework is provided for systematic variation and evaluation of models. The proposed framework considers procedural generation of hypotheses as scientific workflows using a fixed set of functional building blocks, and allows to search for models by seamless evaluation in simulation and real world experiments. Additional contributions of the thesis are related to the agent's causal footprint in sensorimotor time. A probabilistic graphical model is provided, along with an information-theoretic learning algorithm, to discover networks of information flow in sensorimotor data. A generic developmental model, based on real time prediction learning, is presented and discussed on the basis of three different algorithmic variations.
19
Elahi, Haroon. "A Boosted-Window Ensemble." Thesis, Blekinge Tekniska Högskola, Institutionen för datalogi och datorsystemteknik, 2014. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-5658.
Full textAbstract:
Context. The problem of obtaining predictions from stream data involves training on the labeled instances and suggesting the class values for the unseen stream instances. The nature of the data-stream environments makes this task complicated. The large number of instances, the possibility of changes in the data distribution, presence of noise and drifting concepts are just some of the factors that add complexity to the problem. Various supervised-learning algorithms have been designed by putting together efficient data-sampling, ensemble-learning, and incremental-learning methods. The performance of the algorithm is dependent on the chosen methods. This leaves an opportunity to design new supervised-learning algorithms by using different combinations of constructing methods. Objectives. This thesis work proposes a fast and accurate supervised-learning algorithm for performing predictions on the data-streams. This algorithm is called as Boosted-Window Ensemble (BWE), which is invented using the mixture-of-experts technique. BWE uses Sliding Window, Online Boosting and incremental-learning for data-sampling, ensemble-learning, and maintaining a consistent state with the current stream data, respectively. In this regard, a sliding window method is introduced. This method uses partial-updates for sliding the window on the data-stream and is called Partially-Updating Sliding Window (PUSW). The investigation is carried out to compare two variants of sliding window and three different ensemble-learning methods for choosing the superior methods. Methods. The thesis uses experimentation approach for evaluating the Boosted-Window Ensemble (BWE). CPU-time and the Prediction accuracy are used as performance indicators, where CPU-time is the execution time in seconds. The benchmark algorithms include: Accuracy-Updated Ensemble1 (AUE1), Accuracy-Updated Ensemble2 (AUE2), and Accuracy-Weighted Ensemble (AWE). The experiments use nine synthetic and five real-world datasets for generating performance estimates. The Asymptotic Friedman test and the Wilcoxon Signed-Rank test are used for hypothesis testing. The Wilcoxon-Nemenyi-McDonald-Thompson test is used for performing post-hoc analysis. Results. The hypothesis testing suggests that: 1) both for the synthetic and real-wrold datasets, the Boosted Window Ensemble (BWE) has significantly lower CPU-time values than two benchmark algorithms (Accuracy-updated Ensemble1 (AUE1) and Accuracy-weighted Ensemble (AWE). 2) BWE returns similar prediction accuracy as AUE1 and AWE for synthetic datasets. 3) BWE returns similar prediction accuracy as the three benchmark algorithms for the real-world datasets. Conclusions. Experimental results demonstrate that the proposed algorithm can be as accurate as the state-of-the-art benchmark algorithms, while obtaining predictions from the stream data. The results further show that the use of Partially-Updating Sliding Window has resulted in lower CPU-time for BWE as compared with the chunk-based sliding window method used in AUE1, AUE2, and AWE.
20
Murphy, Nicholas John. "An online learning algorithm for technical trading." Master's thesis, Faculty of Science, 2019. http://hdl.handle.net/11427/31048.
Full textAbstract:
We use an adversarial expert based online learning algorithm to learn the optimal parameters required to maximise wealth trading zero-cost portfolio strategies. The learning algorithm is used to determine the relative population dynamics of technical trading strategies that can survive historical back-testing as well as form an overall aggregated portfolio trading strategy from the set of underlying trading strategies implemented on daily and intraday Johannesburg Stock Exchange data. The resulting population time-series are investigated using unsupervised learning for dimensionality reduction and visualisation. A key contribution is that the overall aggregated trading strategies are tested for statistical arbitrage using a novel hypothesis test proposed by Jarrow et al. [31] on both daily sampled and intraday time-scales. The (low frequency) daily sampled strategies fail the arbitrage tests after costs, while the (high frequency) intraday sampled strategies are not falsified as statistical arbitrages after costs. The estimates of trading strategy success, cost of trading and slippage are considered along with an offline benchmark portfolio algorithm for performance comparison. In addition, the algorithms generalisation error is analysed by recovering a probability of back-test overfitting estimate using a nonparametric procedure introduced by Bailey et al. [19]. The work aims to explore and better understand the interplay between different technical trading strategies from a data-informed perspective.
21
Laflamme, Simon M. Eng Massachusetts Institute of Technology. "Online learning algorithm for structural control using magnetorheological actuators." Thesis, Massachusetts Institute of Technology, 2007. http://hdl.handle.net/1721.1/39271.
Full textAbstract:
Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Civil and Environmental Engineering, 2007.Includes bibliographical references (p. 83-84).
Magnetorheological actuators are promising devices for mitigating vibrations because they only require a fraction of energy for a similar performance to active control. Conversely, these semi-active devices have limited maximum forces and are hard to model due to the rheological properties of their fluid. When considering structural control, classical theories necessitate full knowledge of the structural dynamic states and properties most of which can only be estimated when considering large-scale control, which may be difficult or inaccurate for complicated geometries due to the non-linear behaviour of structures. Additionally, most of these theories do not take into account the response delay of the actuators which may result in structural instabilities. To address the problem, learning algorithms using offline learning have been proposed in order to have the structure learn its behaviour, but they can be perceived as unrealistic because earthquake data can hardly be produced to train these schemes. Here, an algorithm using online learning feedback is proposed to address this problem where the structure observes, compares and adapts its performance at each time step, analogous to a child learning his or her motor functions.
(cont.) The algorithm uses a machine learning technique, Gaussian kernels, to prescribe forces upon structural states, where states are evaluated strictly based on displacement and acceleration feedback. The algorithm has been simulated and performances assessed by comparing it with two classical control theories: clipped-optimal and passive controls. The proposed scheme is found to be stable and performs well in mitigating vibrations for a low energy input, but does not perform as well compared to clipped-optimal case. This relative performance would be expected to be better for large-scale structures because of the adaptability of the proposed algorithm.
by Simon Laflamme.
M.Eng.
22
Brégère, Margaux. "Stochastic bandit algorithms for demand side management Simulating Tariff Impact in Electrical Energy Consumption Profiles with Conditional Variational Autoencoders Online Hierarchical Forecasting for Power Consumption Data Target Tracking for Contextual Bandits : Application to Demand Side Management." Thesis, université Paris-Saclay, 2020. http://www.theses.fr/2020UPASM022.
Full textAbstract:
L'électricité se stockant difficilement à grande échelle, l'équilibre entre la production et la consommation doit être rigoureusement maintenu. Une gestion par anticipation de la demande se complexifie avec l'intégration au mix de production des énergies renouvelables intermittentes. Parallèlement, le déploiement des compteurs communicants permet d'envisager un pilotage dynamique de la consommation électrique. Plus concrètement, l'envoi de signaux - tels que des changements du prix de l'électricité – permettrait d'inciter les usagers à moduler leur consommation afin qu'elle s'ajuste au mieux à la production d'électricité. Les algorithmes choisissant ces signaux devront apprendre la réaction des consommateurs face aux envois tout en les optimisant (compromis exploration-exploitation). Notre approche, fondée sur la théorie des bandits, a permis de formaliser ce problème d'apprentissage séquentiel et de proposer un premier algorithme pour piloter la demande électrique d'une population homogène de consommateurs. Une borne supérieure d'ordre T⅔ a été obtenue sur le regret de cet algorithme. Des expériences réalisées sur des données de consommation de foyers soumis à des changements dynamiques du prix de l'électricité illustrent ce résultat théorique. Un jeu de données en « information complète » étant nécessaire pour tester un algorithme de bandits, un simulateur de données de consommation fondé sur les auto-encodeurs variationnels a ensuite été construit. Afin de s'affranchir de l'hypothèse d'homogénéité de la population, une approche pour segmenter les foyers en fonction de leurs habitudes de consommation est aussi proposée. Ces différents travaux sont finalement combinés pour proposer et tester des algorithmes de bandits pour un pilotage personnalisé de la consommation électriqueAs electricity is hard to store, the balance between production and consumption must be strictly maintained. With the integration of intermittent renewable energies into the production mix, the management of the balance becomes complex. At the same time, the deployment of smart meters suggests demand response. More precisely, sending signals - such as changes in the price of electricity - would encourage users to modulate their consumption according to the production of electricity. The algorithms used to choose these signals have to learn consumer reactions and, in the same time, to optimize them (exploration-exploration trade-off). Our approach is based on bandit theory and formalizes this sequential learning problem. We propose a first algorithm to control the electrical demand of a homogeneous population of consumers and offer T⅔ upper bound on its regret. Experiments on a real data set in which price incentives were offered illustrate these theoretical results. As a “full information” dataset is required to test bandit algorithms, a consumption data generator based on variational autoencoders is built. In order to drop the assumption of the population homogeneity, we propose an approach to cluster households according to their consumption profile. These different works are finally combined to propose and test a bandit algorithm for personalized demand side management
23
Peel, Thomas. "Algorithmes de poursuite stochastiques et inégalités de concentration empiriques pour l'apprentissage statistique." Thesis, Aix-Marseille, 2013. http://www.theses.fr/2013AIXM4769/document.
Full textAbstract:
La première partie de cette thèse introduit de nouveaux algorithmes de décomposition parcimonieuse de signaux. Basés sur Matching Pursuit (MP) ils répondent au problème suivant : comment réduire le temps de calcul de l'étape de sélection de MP, souvent très coûteuse. En réponse, nous sous-échantillonnons le dictionnaire à chaque itération, en lignes et en colonnes. Nous montrons que cette approche fondée théoriquement affiche de bons résultats en pratique. Nous proposons ensuite un algorithme itératif de descente de gradient par blocs de coordonnées pour sélectionner des caractéristiques en classification multi-classes. Celui-ci s'appuie sur l'utilisation de codes correcteurs d'erreurs transformant le problème en un problème de représentation parcimonieuse simultanée de signaux. La deuxième partie expose de nouvelles inégalités de concentration empiriques de type Bernstein. En premier, elles concernent la théorie des U-statistiques et sont utilisées pour élaborer des bornes en généralisation dans le cadre d'algorithmes de ranking. Ces bornes tirent parti d'un estimateur de variance pour lequel nous proposons un algorithme de calcul efficace. Ensuite, nous présentons une version empirique de l'inégalité de type Bernstein proposée par Freedman [1975] pour les martingales. Ici encore, la force de notre borne réside dans l'introduction d'un estimateur de variance calculable à partir des données. Cela nous permet de proposer des bornes en généralisation pour l'ensemble des algorithmes d'apprentissage en ligne améliorant l'état de l'art et ouvrant la porte à une nouvelle famille d'algorithmes d'apprentissage tirant parti de cette information empiriqueThe first part of this thesis introduces new algorithms for the sparse encoding of signals. Based on Matching Pursuit (MP) they focus on the following problem : how to reduce the computation time of the selection step of MP. As an answer, we sub-sample the dictionary in line and column at each iteration. We show that this theoretically grounded approach has good empirical performances. We then propose a bloc coordinate gradient descent algorithm for feature selection problems in the multiclass classification setting. Thanks to the use of error-correcting output codes, this task can be seen as a simultaneous sparse encoding of signals problem. The second part exposes new empirical Bernstein inequalities. Firstly, they concern the theory of the U-Statistics and are applied in order to design generalization bounds for ranking algorithms. These bounds take advantage of a variance estimator and we propose an efficient algorithm to compute it. Then, we present an empirical version of the Bernstein type inequality for martingales by Freedman [1975]. Again, the strength of our result lies in the variance estimator computable from the data. This allows us to propose generalization bounds for online learning algorithms which improve the state of the art and pave the way to a new family of learning algorithms taking advantage of this empirical information
24
Provatas, Spyridon. "An Online Machine Learning Algorithm for Heat Load Forecasting in District Heating Systems." Thesis, Blekinge Tekniska Högskola, Institutionen för datalogi och datorsystemteknik, 2014. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-3475.
Full textAbstract:
Context. Heat load forecasting is an important part of district heating optimization. In particular, energy companies aim at minimizing peak boiler usage, optimizing combined heat and power generation and planning base production. To achieve resource efficiency, the energy companies need to estimate how much energy is required to satisfy the market demand. Objectives. We suggest an online machine learning algorithm for heat load forecasting. Online algorithms are increasingly used due to their computational efficiency and their ability to handle changes of the predictive target variable over time. We extend the implementation of online bagging to make it compatible to regression problems and we use the Fast Incremental Model Trees with Drift Detection (FIMT-DD) as the base model. Finally, we implement and incorporate to the algorithm a mechanism that handles missing values, measurement errors and outliers. Methods. To conduct our experiments, we use two machine learning software applications, namely Waikato Environment for Knowledge Analysis (WEKA) and Massive Online Analysis (MOA). The predictive ability of the suggested algorithm is evaluated on operational data from a part of the Karlshamn District Heating network. We investigate two approaches for aggregating the data from the nodes of the network. The algorithm is evaluated on 100 runs using the repeated measures experimental design. A paired T-test is run to test the hypothesis that the the choice of approach does not have a significant effect on the predictive error of the algorithm. Results. The presented algorithm forecasts the heat load with a mean absolute percentage error of 4.77\%. This means that there is a sufficiently accurate estimation of the actual values of the heat load, which can enable heat suppliers to plan and manage more effectively the heat production. Conclusions. Experimental results show that the presented algorithm can be a viable alternative to state-of-the-art algorithms that are used for heat load forecasting. In addition to its predictive ability, it is memory-efficient and can process data in real time. Robust heat load forecasting is an important part of increased system efficiency within district heating, and the presented algorithm provides a concrete foundation for operational usage of online machine learning algorithms within the domain.
25
Barbosa, Haline Pereira de Oliveira, and 5592991791259. "Detecção de Phishing no Twitter Baseada em Algoritmos de Aprendizagem Online." Universidade Federal do Amazonas, 2018. https://tede.ufam.edu.br/handle/tede/6778.
Full textAbstract:
Submitted by Haline Barbosa (halinebarbosa@icomp.ufam.edu.br) on 2018-11-23T12:40:23Z
No. of bitstreams: 2
license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5)
HalinePereiradeOliveiraBarbosa.pdf: 2143170 bytes, checksum: ff7bf1fb1f0781cd5558c12bc7cba05a (MD5)Approved for entry into archive by Secretaria PPGI (secretariappgi@icomp.ufam.edu.br) on 2018-11-23T14:34:32Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) HalinePereiradeOliveiraBarbosa.pdf: 2143170 bytes, checksum: ff7bf1fb1f0781cd5558c12bc7cba05a (MD5)
Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2018-11-23T18:24:02Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) HalinePereiradeOliveiraBarbosa.pdf: 2143170 bytes, checksum: ff7bf1fb1f0781cd5558c12bc7cba05a (MD5)
Made available in DSpace on 2018-11-23T18:24:02Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) HalinePereiradeOliveiraBarbosa.pdf: 2143170 bytes, checksum: ff7bf1fb1f0781cd5558c12bc7cba05a (MD5) Previous issue date: 2018-04-03
Twitter is one of the most used social networks in the world with about 328 million users sharing images, videos, texts and links. Due to the restrictions on message size it is common for tweets to share shortened links to websites, making it impossible to visually identify the URL before knowing what will be displayed. Faced with this scenario, Twitter becomes a means of spreading phishing attacks through malicious links. Phishing is an attack that seeks to obtain personal information like name, CPF, passwords, number of bank accounts and numbers of credit cards. Twitter phishing attack detection systems are usually built using off-line supervised machine learning, where a large amount of data is examined once to induce a single static prediction model. In these systems, the incorporation of new data requires the reconstruction of the prediction model from the processing of the entire database, making this process slow and inefficient. In this work we propose a framework to detect phishing in Twitter. The framework uses supervised online learning, that is, the classifier is updated with each processed tweet and, if it makes a wrong prediction, the model is updated by adapting quickly to the changes with low computational cost, time and maintaining its efficiency in the task of ranking. For this study we evaluated the performance of the online learning algorithms Adaptive Random Forest, Hoeffding Tree, Naive Bayes, Perceptron and Stochastic Gradient Descent. The online Adaptive Random Forest classifier presented 99.8% prequential accuracy in the classification of phishing tweets.
O Twitter é uma das redes sociais mais utilizadas no mundo com cerca de centenas de milhões de usuários compartilhando imagens, vídeos, textos e links. Devido às restrições impostas no tamanho das mensagens é comum que os tweets compartilhem links encurtados para websites impossibilitando a identificação visual prévia da URL antes de saber o que será exibido. Tal problema tornou o Twitter um dos principais meios de disseminação de ataques de phishing através de links maliciosos. Phishing é um ataque que visa obter informações pessoais como nomes, senhas, números de contas bancárias e de cartões de crédito. Em geral, os sistemas de detecção de ataques de phishing projetados para o Twitter são construídos com base em modelos de classificação off-line. Em tais sistemas, um grande volume de dados é examinado uma única vez para induzir em um único modelo de predição estático. Nesses sistemas, a incorporação de novos dados requer a reconstrução do modelo de previsão a partir do processamento de toda a base de dados, tornando esse processo lento e ineficiente. Para solucionar este problema, este trabalho propõe um framework de detecção de phishing no Twitter. O framework utiliza aprendizagem online supervisionada, ou seja, o classificador é atualizado a cada tweet processado e, caso este realize uma predição errada, o modelo é atualizado se adaptando rapidamente às mudanças com baixo custo computacional, tempo e mantendo a sua eficiência na tarefa de classificação. Para este estudo avaliamos o desempenho dos algoritmos de aprendizagem online Adaptive Random Forest, Hoeffding Tree, Naive Bayes, Perceptron e Stochastic Gradient Descent. O classificador online Adaptive Random Forest apresentou acurácia prequential 99,8%, na classificação de tweets de phishing.
26
Farghally, Mohammed Fawzi Seddik. "Visualizing Algorithm Analysis Topics." Diss., Virginia Tech, 2016. http://hdl.handle.net/10919/73539.
Full textAbstract:
Data Structures and Algorithms (DSA) courses are critical for any computer science curriculum. DSA courses emphasize concepts related to procedural dynamics and Algorithm Analysis (AA). These concepts are hard for students to grasp when conveyed using traditional textbook material relying on text and static images. Algorithm Visualizations (AVs) emerged as a technique for conveying DSA concepts using interactive visual representations. Historically, AVs have dealt with portraying algorithm dynamics, and the AV developer community has decades of successful experience with this. But there exist few visualizations to present algorithm analysis concepts. This content is typically still conveyed using text and static images.
We have devised an approach that we term Algorithm Analysis Visualizations (AAVs), capable of conveying AA concepts visually. In AAVs, analysis is presented as a series of slides where each statement of the explanation is connected to visuals that support the sentence. We developed a pool of AAVs targeting the basic concepts of AA. We also developed AAVs for basic sorting algorithms, providing a concrete depiction about how the running time analysis of these algorithms can be calculated.
To evaluate AAVs, we conducted a quasi-experiment across two offerings of CS3114 at Virginia Tech. By analyzing OpenDSA student interaction logs, we found that intervention group students spent significantly more time viewing the material as compared to control group students who used traditional textual content. Intervention group students gave positive feedback regarding the usefulness of AAVs to help them understand the AA concepts presented in the course. In addition, intervention group students demonstrated better performance than control group students on the AA part of the final exam.
The final exam taken by both the control and intervention groups was based on a pilot version of the Algorithm Analysis Concept Inventory (AACI) that was developed to target fundamental AA concepts and probe students' misconceptions about these concepts. The pilot AACI was developed using a Delphi process involving a group of DSA instructors, and was shown to be a valid and reliable instrument to gauge students' understanding of the basic AA topics.Ph. D.
27
Liakopoulos, Nikolaos. "Machine Learning Techniques for Online Resource Allocation in Wireless Networks." Electronic Thesis or Diss., Sorbonne université, 2019. http://www.theses.fr/2019SORUS529.
Full textAbstract:
Traditionnellement, l'optimisation du réseau est utilisée afin de fournir de bonnes options de configuration basées sur des modèles mathématiques et des hypothèses statistiques. La tendance actuelle en ce qui concerne des problèmes de réseau consiste à exploiter le pouvoir des données afin d’extraire des modèles et gérer les incertitudes. Cette thèse, propose des cadres algorithmiques pour les réseaux sans fil, basés à la fois sur l'optimisation classique ou des données et sur l'apprentissage automatique. Nous ciblons deux cas, l'association d'utilisateurs et la réservation de ressources en nuages. Dans le premier cas, l'approche de base pour l'association d'utilisateurs qui consiste à connecter des périphériques sans fil à la station de base fournissant le signal le plus puissant, conduit à des configurations inefficaces. De prime abord, nous étudions, l’association d’utilisateurs distribués avec des garanties de qualité de service prioritaires, puis nous prenons en considération l’équilibrage de charge centralisé évolutif, basé sur le transport optimal par calcul et enfin nous prenons l’exemple d’une association d’utilisateurs robuste basée sur la prédiction de trafic approximative.Ensuite, nous développons un nouveau cadre pour la réservation de ressources en nuages dans les scénarios les plus défavorables, dans lequel la demande est gérée par un adversaire. Nous proposons des politiques «no regret» et nous garantissons une faisabilité asymptotique des contraintes budgétaires. Nous développons un cadre général, pour des problèmes d’optimisation, convexe en ligne (OCO) avec contraintes budgétaires à long terme complétant les résultats de la littérature récenteTraditionally, network optimization is used to provide good configurations in real network system problems based on mathematical models and statistical assumptions. Recently, this paradigm is evolving, fueled by an explosion of availability of data. The modern trend in networking problems is to tap into the power of data to extract models and deal with uncertainty. This thesis proposes algorithmic frameworks for wireless networks, based both on classical or data-driven optimization and machine learning. We target two use cases, user association and cloud resource reservation.The baseline approach for user association, connecting wireless devices to the base station that provides the strongest signal, leads to very inefficient configurations even in current wireless networks. We focus on tailoring user association based on resource efficiency and service requirement satisfaction, depending on the underlying network demand. We first study distributed user association with priority QoS guarantees, then scalable centralized load balancing based on computational optimal transport and finally robust user association based on approximate traffic prediction.Moving to the topic of cloud resource reservation, we develop a novel framework for resource reservation in worst-case scenaria, where the demand is engineered by an adversary aiming to harm our performance. We provide policies that have ``no regret'' and guarantee asymptotic feasibility in budget constraints under such workloads. More importantly we expand to a general framework for online convex optimization (OCO) problems with long term budget constraints complementing the results of recent literature in OCO
28
Holm, Raven R. "Natural language processing of online propaganda as a means of passively monitoring an adversarial ideology." Thesis, Monterey, California: Naval Postgraduate School, 2017. http://hdl.handle.net/10945/52993.
Full textAbstract:
Approved for public release; distribution is unlimitedReissued 30 May 2017 with Second Reader’s non-NPS affiliation added to title page.
Online propaganda embodies a potent new form of warfare; one that extends the strategic reach of our adversaries and overwhelms analysts. Foreign organizations have effectively leveraged an online presence to influence elections and distance-recruit. The Islamic State has also shown proficiency in outsourcing violence, proving that propaganda can enable an organization to wage physical war at very little cost and without the resources traditionally required. To augment new counter foreign propaganda initiatives, this thesis presents a pipeline for defining, detecting and monitoring ideology in text. A corpus of 3,049 modern online texts was assembled and two classifiers were created: one for detecting authorship and another for detecting ideology. The classifiers demonstrated 92.70% recall and 95.84% precision in detecting authorship, and detected ideological content with 76.53% recall and 95.61% precision. Both classifiers were combined to simulate how an ideology can be detected and how its composition could be passively monitored across time. Implementation of such a system could conserve manpower in the intelligence community and add a new dimension to analysis. Although this pipeline makes presumptions about the quality and integrity of input, it is a novel contribution to the fields of Natural Language Processing and Information Warfare.
Lieutenant, United States Coast Guard
29
Duarte, Kevin. "Aide à la décision médicale et télémédecine dans le suivi de l’insuffisance cardiaque." Thesis, Université de Lorraine, 2018. http://www.theses.fr/2018LORR0283/document.
Full textAbstract:
Cette thèse s’inscrit dans le cadre du projet "Prendre votre cœur en mains" visant à développer un dispositif médical d’aide à la prescription médicamenteuse pour les insuffisants cardiaques. Dans une première partie, une étude a été menée afin de mettre en évidence la valeur pronostique d’une estimation du volume plasmatique ou de ses variations pour la prédiction des événements cardiovasculaires majeurs à court terme. Deux règles de classification ont été utilisées, la régression logistique et l’analyse discriminante linéaire, chacune précédée d’une phase de sélection pas à pas des variables. Trois indices permettant de mesurer l’amélioration de la capacité de discrimination par ajout du biomarqueur d’intérêt ont été utilisés. Dans une seconde partie, afin d’identifier les patients à risque de décéder ou d’être hospitalisé pour progression de l’insuffisance cardiaque à court terme, un score d’événement a été construit par une méthode d’ensemble, en utilisant deux règles de classification, la régression logistique et l’analyse discriminante linéaire de données mixtes, des échantillons bootstrap et en sélectionnant aléatoirement les prédicteurs. Nous définissons une mesure du risque d’événement par un odds-ratio et une mesure de l’importance des variables et des groupes de variables. Nous montrons une propriété de l’analyse discriminante linéaire de données mixtes. Cette méthode peut être mise en œuvre dans le cadre de l’apprentissage en ligne, en utilisant des algorithmes de gradient stochastique pour mettre à jour en ligne les prédicteurs. Nous traitons le problème de la régression linéaire multidimensionnelle séquentielle, en particulier dans le cas d’un flux de données, en utilisant un processus d’approximation stochastique. Pour éviter le phénomène d’explosion numérique et réduire le temps de calcul pour prendre en compte un maximum de données entrantes, nous proposons d’utiliser un processus avec des données standardisées en ligne au lieu des données brutes et d’utiliser plusieurs observations à chaque étape ou toutes les observations jusqu’à l’étape courante sans avoir à les stocker. Nous définissons trois processus et en étudions la convergence presque sûre, un avec un pas variable, un processus moyennisé avec un pas constant, un processus avec un pas constant ou variable et l’utilisation de toutes les observations jusqu’à l’étape courante. Ces processus sont comparés à des processus classiques sur 11 jeux de données. Le troisième processus à pas constant est celui qui donne généralement les meilleurs résultatsThis thesis is part of the "Handle your heart" project aimed at developing a drug prescription assistance device for heart failure patients. In a first part, a study was conducted to highlight the prognostic value of an estimation of plasma volume or its variations for predicting major short-term cardiovascular events. Two classification rules were used, logistic regression and linear discriminant analysis, each preceded by a stepwise variable selection. Three indices to measure the improvement in discrimination ability by adding the biomarker of interest were used. In a second part, in order to identify patients at short-term risk of dying or being hospitalized for progression of heart failure, a short-term event risk score was constructed by an ensemble method, two classification rules, logistic regression and linear discriminant analysis of mixed data, bootstrap samples, and by randomly selecting predictors. We define an event risk measure by an odds-ratio and a measure of the importance of variables and groups of variables using standardized coefficients. We show a property of linear discriminant analysis of mixed data. This methodology for constructing a risk score can be implemented as part of online learning, using stochastic gradient algorithms to update online the predictors. We address the problem of sequential multidimensional linear regression, particularly in the case of a data stream, using a stochastic approximation process. To avoid the phenomenon of numerical explosion which can be encountered and to reduce the computing time in order to take into account a maximum of arriving data, we propose to use a process with online standardized data instead of raw data and to use of several observations per step or all observations until the current step. We define three processes and study their almost sure convergence, one with a variable step-size, an averaged process with a constant step-size, a process with a constant or variable step-size and the use of all observations until the current step without storing them. These processes are compared to classical processes on 11 datasets. The third defined process with constant step-size typically yields the best results
30
Raykhel, Ilya Igorevitch. "Real-Time Automatic Price Prediction for eBay Online Trading." BYU ScholarsArchive, 2008. https://scholarsarchive.byu.edu/etd/1631.
Full textAbstract:
While Machine Learning is one of the most popular research areas in Computer Science, there are still only a few deployed applications intended for use by the general public. We have developed an exemplary application that can be directly applied to eBay trading. Our system predicts how much an item would sell for on eBay based on that item's attributes. We ran our experiments on the eBay laptop category, with prior trades used as training data. The system implements a feature-weighted k-Nearest Neighbor algorithm, using genetic algorithms to determine feature weights. Our results demonstrate an average prediction error of 16%; we have also shown that this application greatly reduces the time a reseller would need to spend on trading activities, since the bulk of market research is now done automatically with the help of the learned model.
31
Labernia, Fabien. "Algorithmes efficaces pour l’apprentissage de réseaux de préférences conditionnelles à partir de données bruitées." Thesis, Paris Sciences et Lettres (ComUE), 2018. http://www.theses.fr/2018PSLED018/document.
Full textAbstract:
La croissance exponentielle des données personnelles, et leur mise à disposition sur la toile, a motivé l’émergence d’algorithmes d’apprentissage de préférences à des fins de recommandation, ou d’aide à la décision. Les réseaux de préférences conditionnelles (CP-nets) fournissent une structure compacte et intuitive pour la représentation de telles préférences. Cependant, leur nature combinatoire rend leur apprentissage difficile : comment apprendre efficacement un CP-net au sein d’un milieu bruité, tout en supportant le passage à l’échelle ?Notre réponse prend la forme de deux algorithmes d’apprentissage dont l’efficacité est soutenue par de multiples expériences effectuées sur des données réelles et synthétiques.Le premier algorithme se base sur des requêtes posées à des utilisateurs, tout en prenant en compte leurs divergences d’opinions. Le deuxième algorithme, composé d’une version hors ligne et en ligne, effectue une analyse statistique des préférences reçues et potentiellement bruitées. La borne de McDiarmid est en outre utilisée afin de garantir un apprentissage en ligne efficaceThe rapid growth of personal web data has motivated the emergence of learning algorithms well suited to capture users’ preferences. Among preference representation formalisms, conditional preference networks (CP-nets) have proven to be effective due to their compact and explainable structure. However, their learning is difficult due to their combinatorial nature.In this thesis, we tackle the problem of learning CP-nets from corrupted large datasets. Three new algorithms are introduced and studied on both synthetic and real datasets.The first algorithm is based on query learning and considers the contradictions between multiple users’ preferences by searching in a principled way the variables that affect the preferences. The second algorithm relies on information-theoretic measures defined over the induced preference rules, which allow us to deal with corrupted data. An online version of this algorithm is also provided, by exploiting the McDiarmid's bound to define an asymptotically optimal decision criterion for selecting the best conditioned variable and hence allowing to deal with possibly infinite data streams
32
Cayuela, Rafols Marc. "Algorithmic Study on Prediction with Expert Advice : Study of 3 novel paradigms with Grouped Experts." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-254344.
Full textAbstract:
The main work for this thesis has been a thorough study of the novel Prediction with Partially Monitored Grouped Expert Advice and Side Information paradigm. This is newly proposed in this thesis, and it extends the widely studied Prediction with Expert Advice paradigm. The extension is based on two assumptions and one restriction that modify the original problem. The first assumption, Grouped, presumes that the experts are structured into groups. The second assumption, Side Information, introduces additional information that can be used to timely relate predictions with groups. Finally, the restriction, Partially Monitored, imposes that the groups’ predictions are only known for one group at a time. The study of this paradigm includes the design of a complete prediction algorithm, the proof of a theoretical bound of the worse-case cumulative regret for such algorithm, and an experimental evaluation of the algorithm (proving the existence of cases where this paradigm outperforms Prediction with Expert Advice). Furthermore, since the development of the algorithm is constructive, it allows to easily build two additional prediction algorithms for the Prediction with Grouped Expert Advice and Prediction with Grouped Expert Advice and Side Information paradigms. Therefore, this thesis presents three novel prediction algorithms, with corresponding regret bounds, and a comparative experimental evaluation including the original Prediction with Expert Advice paradigm.Huvudarbetet för den här avhandlingen har varit en grundlig studie av den nya Prediction with Partially Monitored Grouped Expert Advice and Side Information paradigmet. Detta är nyligen föreslagit i denna avhandling, och det utökar det brett studerade Prediction with Expert Advice paradigmet. Förlängningen baseras på två antaganden och en begränsning som ändrar det ursprungliga problemet. Det första antagandet, Grouped, förutsätter att experterna är inbyggda i grupper. Det andra antagandet, Side Information, introducerar ytterligare information som kan användas för att i tid relatera förutsägelser med grupper. Slutligen innebär begränsningen, Partially Monitored, att gruppens förutsägelser endast är kända för en grupp i taget. Studien av detta paradigm innefattar utformningen av en komplett förutsägelsesalgoritm, beviset på en teoretisk bindning till det sämre fallet kumulativa ånger för en sådan algoritm och en experimentell utvärdering av algoritmen (bevisar förekomsten av fall där detta paradigm överträffar Prediction with Expert Advice). Eftersom algoritmens utveckling är konstruktiv tillåter den dessutom att enkelt bygga två ytterligare prediksionsalgoritmer för Prediction with Grouped Expert Advice och Prediction with Grouped Expert Advice and Side Information paradigmer. Därför presenterar denna avhandling tre nya prediktionsalgoritmer med motsvarande ångergränser och en jämförande experimentell utvärdering inklusive det ursprungliga Prediction with Expert Advice paradigmet.
33
SILVA, Márcio Eduardo Gonçalves. "Algoritmos da Família LMS para a Solução Aproximada da HJB em Projetos Online de Controle Ótimo Discreto Multivariável e Aprendizado por Reforço." Universidade Federal do Maranhão, 2014. http://tedebc.ufma.br:8080/jspui/handle/tede/1891.
Full textAbstract:
Submitted by Maria Aparecida (cidazen@gmail.com) on 2017-09-04T13:10:41Z
No. of bitstreams: 1
Marcio Eduardo.pdf: 7939176 bytes, checksum: 3b90c4b32aeabafd3b87e4f3c36d2ed6 (MD5)Made available in DSpace on 2017-09-04T13:10:41Z (GMT). No. of bitstreams: 1 Marcio Eduardo.pdf: 7939176 bytes, checksum: 3b90c4b32aeabafd3b87e4f3c36d2ed6 (MD5) Previous issue date: 2014-08-21
The technique of linear control based on the minimization of a quadratic performance index using the second method of Lyapunov to guarantee the stability of the system, if this is controllable and observable. however, this technique is inevitably necessary to find the solution of the HJB or Riccati equation. The control system design online need, real time, to adjust your feedback gain to maintain a certain dynamic, it requires the calculation of the Riccati equation solution in each sampling generating a large computational load that can derail its implementation. This work shows an intelligent control system design that meets the optimal or suboptimal control action from the sensory data of process states and the instantaneous cost observed after each state transition. To find this optimal control action or policy, the approximate dynamic programming and adaptive critics are used, based on the parameterizations given by the problem of linear quadratic regulator (LQR), but without explicitly solving the associated Riccati equation. More specifically, the LQR problem is solved by four different methods which are the Dynamic Programming Heuristic, the Dual Heuristic Dynamic Programming, Action Dependent Dynamic Programming Heuristic and Action Dependent Dual Heuristic Dynamic Programming algorithms. However, these algorithms depend on knowledge of the value functions to derive the optimal control actions. These value functions with known structures have their parameters estimated using the least mean square family and Recursive Least Squares algorithms. Two processes that have the Markov property were used in the computational validation of the algorithms adaptive critics implemented, one corresponds to the longitudinal dynamics of an aircraft and the other to an electrical circuit.
A técnica de controle linear baseado na minimização de um índices de desempenho quadrático utilizando o segundo método de Liapunov garante a estabilidade do sistema, se este for controlável e observável. Por outro lado, nessa técnica inexoravelmente é necessário encontrar a solução da Equação Hamilton-Jacobi-Bellman (HJB) ou Riccati. Em projeto de sistema de controle online que necessita, em tempo real, alterar seus ganhos de retroação para manter uma certa dinâmica, impõe o cálculo da solução da equação de Riccati em cada instante de amostragem gerando uma grande carga computacional que pode inviabilizar sua implementação. Neste trabalho, mostra-se o projeto de um sistema de controle inteligente que encontra a ação de controle ótima ou subótima a partir de dados sensoriais dos estados do processo e do custo instantâneo observados após cada transição de estado. Para encontrar essa ação de controle ou política ótima, a programação dinâmica aproximada ou críticos adaptativos são utilizados, tendo como base as parametrizações dado pelo problema do regulador linear quadrático (LQR), mas sem resolver explicitamente a equação de Riccati associada. Mais especificamente, o problema do LQR é resolvido por quatro métodos distintos que são os algoritmos de Programação Dinâmica Heurística, a Programação Dinâmica Heurística Dual, a Programação Dinâmica Heurística Dependente de Ação e a Programação Dinâmica Heurística Dual Dependente de Ação. Entretanto, esses algoritmos dependem do conhecimento das funções valor para, assim, derivar as ações de controle ótimas. Essas funções valor com estruturas conhecidas tem seus parâmetros estimados utilizando os algoritmos da família dos mínimos quadrados médios e o algoritmo de Mínimos Quadrados Recursivo. Dois processos que obedecem à propriedade de Markov foram empregados na validação computacional dos algoritmos críticos adaptativos, um corresponde à dinâmica longitudinal de uma aeronave e o outro à de um circuito elétrico.
34
Moscu, Mircea. "Inférence distribuée de topologie de graphe à partir de flots de données." Thesis, Université Côte d'Azur, 2020. http://www.theses.fr/2020COAZ4081.
Full textAbstract:
La deuxième décennie du millénaire actuel peut être résumée en une courte phrase : l'essor des données. Le nombre de sources de données s'est multiplié : du streaming audio-vidéo aux réseaux sociaux et à l'Internet des Objets, en passant par les montres intelligentes, les équipements industriels et les véhicules personnels, pour n'en citer que quelques-unes. Le plus souvent, ces sources forment des réseaux afin d'échanger des informations. En conséquence directe, le domaine du Traitement de Signal sur Graphe a prospéré et a évolué. Son but : traiter et donner un sens à tout le déluge de données environnant. Dans ce contexte, le but principal de cette thèse est de développer des méthodes et des algorithmes capables d'utiliser des flots de données, de manière distribuée, afin d'inférer les réseaux sous-jacents qui relient ces flots. Ensuite, ces topologies de réseau estimées peuvent être utilisées avec des outils développés pour le Traitement de Signal sur Graphe afin de traiter et d'analyser les données supportées par des graphes. Après une brève introduction suivie d'exemples motivants, nous développons et proposons d'abord un algorithme en ligne, distribué et adaptatif pour l'inférence de topologies de graphes pour les flots de données qui sont linéairement dépendants. Une analyse de la méthode s'ensuit, afin d'établir des relations entre les performances et les paramètres nécessaires à l'algorithme. Nous menons ensuite une série d'expériences afin de valider l'analyse et de comparer ses performances avec celles d'une autre méthode proposée dans la littérature. La contribution suivante est un algorithme doté des mêmes capacités en ligne, distribuées et adaptatives, mais adapté à l'inférence de liens entre des données qui interagissent de manière non-linéaire. À ce titre, nous proposons un modèle additif simple mais efficace qui utilise l'usine du noyau reproduisant afin de modéliser lesdites non-linéarités. Les résultats de son analyse sont convaincants, tandis que les expériences menées sur des données biomédicales donnent des réseaux estimés qui présentent un comportement prédit par la littérature médicale. Enfin, une troisième proposition d'algorithme est faite, qui vise à améliorer le modèle non-linéaire en lui permettant d'échapper aux contraintes induites par l'additivité. Ainsi, le nouveau modèle proposé est aussi général que possible, et utilise une manière naturelle et intuitive d'imposer la parcimonie des liens, basée sur le concept de dérivés partiels. Nous analysons également l'algorithme proposé, afin d'établir les conditions de stabilité et les relations entre ses paramètres et ses performances. Une série d'expériences est menée, montrant comment le modèle général est capable de mieux saisir les liens non-linéaires entre les données, tandis que les réseaux estimés se comportent de manière cohérente avec les estimations précédentesThe second decade of the current millennium can be summarized in one short phrase: the advent of data. There has been a surge in the number of data sources: from audio-video streaming, social networks and the Internet of Things, to smartwatches, industrial equipment and personal vehicles, just to name a few. More often than not, these sources form networks in order to exchange information. As a direct consequence, the field of Graph Signal Processing has been thriving and evolving. Its aim: process and make sense of all the surrounding data deluge.In this context, the main goal of this thesis is developing methods and algorithms capable of using data streams, in a distributed fashion, in order to infer the underlying networks that link these streams. Then, these estimated network topologies can be used with tools developed for Graph Signal Processing in order to process and analyze data supported by graphs. After a brief introduction followed by motivating examples, we first develop and propose an online, distributed and adaptive algorithm for graph topology inference for data streams which are linearly dependent. An analysis of the method ensues, in order to establish relations between performance and the input parameters of the algorithm. We then run a set of experiments in order to validate the analysis, as well as compare its performance with that of another proposed method of the literature.The next contribution is in the shape of an algorithm endowed with the same online, distributed and adaptive capacities, but adapted to inferring links between data that interact non-linearly. As such, we propose a simple yet effective additive model which makes use of the reproducing kernel machinery in order to model said nonlinearities. The results if its analysis are convincing, while experiments ran on biomedical data yield estimated networks which exhibit behavior predicted by medical literature.Finally, a third algorithm proposition is made, which aims to improve the nonlinear model by allowing it to escape the constraints induced by additivity. As such, the newly proposed model is as general as possible, and makes use of a natural and intuitive manner of imposing link sparsity, based on the concept of partial derivatives. We analyze this proposed algorithm as well, in order to establish stability conditions and relations between its parameters and its performance. A set of experiments are ran, showcasing how the general model is able to better capture nonlinear links in the data, while the estimated networks behave coherently with previous estimates
35
RÊGO, Patrícia Helena Moraes. "Aprendizagem por Reforço e Programação Dinâmica Aproximada para Controle Ótimo: Uma Abordagem para o Projeto Online do Regulador Linear Quadrático Discreto com Programação Dinâmica Heurística Dependente de Estado e Ação." Universidade Federal do Maranhão, 2014. http://tedebc.ufma.br:8080/jspui/handle/tede/1879.
Full textAbstract:
Submitted by Maria Aparecida (cidazen@gmail.com) on 2017-08-30T15:33:12Z
No. of bitstreams: 1
Patricia Helena.pdf: 11110405 bytes, checksum: ca1f067231658f897d84b86181dbf1b9 (MD5)Made available in DSpace on 2017-08-30T15:33:12Z (GMT). No. of bitstreams: 1 Patricia Helena.pdf: 11110405 bytes, checksum: ca1f067231658f897d84b86181dbf1b9 (MD5) Previous issue date: 2014-07-24
In this thesis a proposal of an uni ed approach of dynamic programming, reinforcement learning and function approximation theories aiming at the development of methods and algorithms for design of optimal control systems is presented. This approach is presented in the approximate dynamic programming context that allows approximating the optimal feedback solution as to reduce the computational complexity associated to the conventional dynamic programming methods for optimal control of multivariable systems. Speci cally, in the state and action dependent heuristic dynamic programming framework, this proposal is oriented for the development of online approximated solutions, numerically stable, of the Riccati-type Hamilton-Jacobi-Bellman equation associated to the discrete linear quadratic regulator problem which is based on a formulation that combines value function estimates by means of a RLS (Recursive Least-Squares) structure, temporal di erences and policy improvements. The development of the proposed methodologies, in this work, is focused mainly on the UDU T factorization that is inserted in this framework to improve the RLS estimation process of optimal decision policies of the discrete linear quadratic regulator, by circumventing convergence and numerical stability problems related to the covariance matrix ill-conditioning of the RLS approach.
Apresenta-se nesta tese uma proposta de uma abordagem uni cada de teorias de programação dinâmica, aprendizagem por reforço e aproximação de função que tem por objetivo o desenvolvimento de métodos e algoritmos para projeto online de sistemas de controle ótimo. Esta abordagem é apresentada no contexto de programação dinâmica aproximada que permite aproximar a solução de realimentação ótima de modo a reduzir a complexidade computacional associada com métodos convencionais de programação dinâmica para controle ótimo de sistemas multivariáveis. Especi camente, no quadro de programação dinâmica heurística e programação dinâmica heurística dependente de ação, esta proposta é orientada para o desenvolvimento de soluções aproximadas online, numericamente estáveis, da equação de Hamilton-Jacobi-Bellman do tipo Riccati associada ao problema do regulador linear quadrático discreto que tem por base uma formulação que combina estimativas da função valor por meio de uma estrutura RLS (do inglês Recursive Least-Squares), diferenças temporais e melhorias de política. O desenvolvimento das metodologias propostas, neste trabalho, tem seu foco principal voltado para a fatoração UDU T que é inserida neste quadro para melhorar o processo de estimação RLS de políticas de decisão ótimas do regulador linear quadrá- tico discreto, contornando-se problemas de convergência e estabilidade numérica relacionados com o mal condicionamento da matriz de covariância da abordagem RLS.
36
Bubeck, Sébastien. "JEUX DE BANDITS ET FONDATIONS DU CLUSTERING." Phd thesis, Université des Sciences et Technologie de Lille - Lille I, 2010. http://tel.archives-ouvertes.fr/tel-00845565.
Full textAbstract:
Ce travail de thèse s'inscrit dans le domaine du machine learning et concerne plus particulièrement les sous-catégories de l'optimisation stochastique, du online learning et du clustering. Ces sous-domaines existent depuis plusieurs décennies mais ils ont tous reçu un éclairage différent au cours de ces dernières années. Notamment, les jeux de bandits offrent aujourd'hui un cadre commun pour l'optimisation stochastique et l'online learning. Ce point de vue conduit a de nombreuses extensions du jeu de base. C'est sur l'étude mathématique de ces jeux que se concentre la première partie de cette thèse. La seconde partie est quant à elle dédiée au clustering et plus particulièrement à deux notions importantes: la consistance asymptotique des algorithmes et la stabilité comme méthode de sélection de modèles.
37
Harrington, Edward. "Aspects of Online Learning." Phd thesis, 2004. http://hdl.handle.net/1885/47147.
Full textAbstract:
Online learning algorithms have several key advantages compared to their batch learning algorithm counterparts. This thesis investigates several online learning algorithms and their application. The thesis has an underlying theme of the idea of combining several simple algorithms to give better performance. In this thesis we investigate: combining weights, combining hypothesis, and (sort of) hierarchical combining.¶ ...
38
Robards, Matthew Walters. "Online learning algorithms for reinforcement learning with function approximation." Phd thesis, 2011. http://hdl.handle.net/1885/150825.
Full textAbstract:
Reinforcement learning deals with the problem of sequential decision making in uncertain stochastic environments. In this thesis I deal with agents who attempt to solve the reinforcement learning problem online and in real-time. This presents experimental challenges for which I introduce novel kernelised algorithms. Kernel algorithms are very useful in reinforcement learning settings as they enable learning in situations where a very high-dimensional or hand engineered feature vector would otherwise be required. Furthermore, I attempt to address the theoretical challenges which arise from online on-policy algorithms, for which I introduce a type of analysis which is novel (and useful) to reinforcement learning in its lack of restrictive assumptions on the behaviour policy. I will introduce three novel algorithms attempting to advance the areas of kernel, empirical and theoretical reinforcement learning. The first of these algorithms presents a kernel extension of SARSA for its empirical properties - namely its incorporation of eligibility traces with sparse kernel algorithms. I then present a model-free/model-based ensemble which use gradient based methods for online learning. I present them with regret analysis which enables an analysis of the value functions learned with no probabilistic assumptions, and hence no assumptions on the behaviour policy. Along the way I also make a novel "sub-contribution", namely non-squared loss functions for reinforcement learning. The use of different loss functions constitutes a running theme through the algorithms I introduce, as I show that various non-traditional (to reinforcement learning) loss functions can be useful for both efficiency of the algorithm, and for accuracy by ensuring smooth function approximations. I present thorough experimental and theoretical analyses along the way.
39
Lakshmanan, K. "Online Learning and Simulation Based Algorithms for Stochastic Optimization." Thesis, 2012. http://hdl.handle.net/2005/3245.
Full textAbstract:
In many optimization problems, the relationship between the objective and parameters is not known. The objective function itself may be stochastic such as a long-run average over some random cost samples. In such cases finding the gradient of the objective is not possible. It is in this setting that stochastic approximation algorithms are used. These algorithms use some estimates of the gradient and are stochastic in nature. Amongst gradient estimation techniques, Simultaneous Perturbation Stochastic Approximation (SPSA) and Smoothed Functional(SF) scheme are widely used. In this thesis we have proposed a novel multi-time scale quasi-Newton based smoothed functional (QN-SF) algorithm for unconstrained as well as constrained optimization. The algorithm uses the smoothed functional scheme for estimating the gradient and the quasi-Newton method to solve the optimization problem. The algorithm is shown to converge with probability one.
We have also provided here experimental results on the problem of optimal routing in a multi-stage network of queues. Policies like Join the Shortest Queue or Least Work Left assume knowledge of the queue length values that can change rapidly or hard to estimate. If the only information available is the expected end-to-end delay as with our case, such policies cannot be used. The QN-SF based probabilistic routing algorithm uses only the total end-to-end delay for tuning the probabilities. We observe from the experiments that the QN-SF algorithm has better performance than the gradient and Jacobi versions of Newton based smoothed functional algorithms. Next we consider constrained routing in a similar queueing network. We extend the QN-SF algorithm to this case. We study the convergence behavior of the algorithm and observe that the constraints are satisfied at the point of convergence. We provide experimental results for the constrained routing setup as well.
Next we study reinforcement learning algorithms which are useful for solving Markov Decision Process(MDP) when the precise information on transition probabilities is not known. When the state, and action sets are very large, it is not possible to store all the state-action tuples. In such cases, function approximators like neural networks have been used. The popular Q-learning algorithm is known to diverge when used with linear function approximation due to the ’off-policy’ problem. Hence developing stable learning algorithms when used with function approximation is an important problem.
We present in this thesis a variant of Q-learning with linear function approximation that is based on two-timescale stochastic approximation. The Q-value parameters for a given policy in our algorithm are updated on the slower timescale while the policy parameters themselves are updated on the faster scale. We perform a gradient search in the space of policy parameters. Since the objective function and hence the gradient are not analytically known, we employ the efficient one-simulation simultaneous perturbation stochastic approximation(SPSA) gradient estimates that employ Hadamard matrix based deterministic perturbations. Our algorithm has the advantage that, unlike Q-learning, it does not suffer from high oscillations due to the off-policy problem when using function approximators. Whereas it is difficult to prove convergence of regular Q-learning with linear function approximation because of the off-policy problem, we prove that our algorithm which is on-policy is convergent. Numerical results on a multi-stage stochastic shortest path problem show that our algorithm exhibits significantly better performance and is more robust as compared to Q-learning. Future work would be to compare it with other policy-based reinforcement learning algorithms.
Finally, we develop an online actor-critic reinforcement learning algorithm with function approximation for a problem of control under inequality constraints. We consider the long-run average cost Markov decision process(MDP) framework in which both the objective and the constraint functions are suitable policy-dependent long-run averages of certain sample path functions. The Lagrange multiplier method is used to handle the inequality constraints. We prove the asymptotic almost sure convergence of our algorithm to a locally optimal solution. We also provide the results of numerical experiments on a problem of routing in a multistage queueing network with constraints on long-run average queue lengths. We observe that our algorithm exhibits good performance on this setting and converges to a feasible point.
40
賴翔偉. "Online Learning Algorithms based on Bayesian IRT models." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/9gn772.
Full textAbstract:
碩士國立政治大學
統計學系
106
In this paper, we present two types of online learning algorithms--statical and dynamical--to capture users’ and items’ latent traits’ information through online product rating data in a real-time manner. The statical one extends Weng and Coad (2018)’s deterministic moment-matching method by adding priors to cutpoints, and the dynamical one extends the statical one with the dynamical ideas adopted in Graepel et al. (2010) for taking users’ and items’ time-dependent latent traits into account. Both learning algorithms are designed for the Bayesian ordinal IRT model proposed by Ho and Quinn (2008). Through experiments, we have verified two things: First, updating cutpoints sequentially produces better results. Second, statical learning’s computational time is almost twice as less as dynamical learning’s, but dynamical learning can slightly outperform statical learning under some configurations. At the end of the paper, we give some useful configurations for setting up the priors of the latent variables of Ho and Quinn’s ordinal IRT model.
41
Chiu, Chien-Jung, and 邱健榮. "Design of an intelligent control system with online learning algorithms." Thesis, 2012. http://ndltd.ncl.edu.tw/handle/23355470340408394886.
Full textAbstract:
博士國立中央大學
電機工程研究所
100
The traditional control system designs are always based on the system dynamic equations; however, it is difficult to be described as the plants are too complex. This dissertation proposes several intelligent control systems based on the adaptive control, sliding-mode control and neural network control technologies. For the adaptive proportional integral derivative (PID) controller design, the adaptive PID controller can automatically tune the controller gain factors based on the gradient descent method. For the adaptive neural network controllers design, a recurrent-wavelet-neural-network-based adaptive control, RBF-neural-network-based adaptive control and fuzzy-wavelet-neural-network-based adaptive control methods are proposed. In these control system designs, an online parameter tuning methodology, using the gradient descent method or the Lyapunov stability theorem, is developed to increase the learning capability and to guarantee the system’s stability. Moreover, a PID type adaptation tuning mechanism is derived to speed up the convergence of the tracking error and controller parameters. Furthermore, the dynamic-sliding-mode-neural-network-based adaptive control design method is developed with dynamic learning rate which is proposed to reduce the chattering phenomenon. Finally, the developed control system design methods are applied to some control system applications, such as induction servomotor system, brushless DC motor system, chaotic system and chaotic synchronization system. The simulation and experimental results have demonstrated that the effectiveness of the proposed design methods.
42
"Learning with Attributed Networks: Algorithms and Applications." Doctoral diss., 2019. http://hdl.handle.net/2286/R.I.54837.
Full textAbstract:
abstract: Attributes - that delineating the properties of data, and connections - that describing the dependencies of data, are two essential components to characterize most real-world phenomena. The synergy between these two principal elements renders a unique data representation - the attributed networks. In many cases, people are inundated with vast amounts of data that can be structured into attributed networks, and their use has been attractive to researchers and practitioners in different disciplines. For example, in social media, users interact with each other and also post personalized content; in scientific collaboration, researchers cooperate and are distinct from peers by their unique research interests; in complex diseases studies, rich gene expression complements to the gene-regulatory networks. Clearly, attributed networks are ubiquitous and form a critical component of modern information infrastructure. To gain deep insights from such networks, it requires a fundamental understanding of their unique characteristics and be aware of the related computational challenges.
My dissertation research aims to develop a suite of novel learning algorithms to understand, characterize, and gain actionable insights from attributed networks, to benefit high-impact real-world applications. In the first part of this dissertation, I mainly focus on developing learning algorithms for attributed networks in a static environment at two different levels: (i) attribute level - by designing feature selection algorithms to find high-quality features that are tightly correlated with the network topology; and (ii) node level - by presenting network embedding algorithms to learn discriminative node embeddings by preserving node proximity w.r.t. network topology structure and node attribute similarity. As changes are essential components of attributed networks and the results of learning algorithms will become stale over time, in the second part of this dissertation, I propose a family of online algorithms for attributed networks in a dynamic environment to continuously update the learning results on the fly. In fact, developing application-aware learning algorithms is more desired with a clear understanding of the application domains and their unique intents. As such, in the third part of this dissertation, I am also committed to advancing real-world applications on attributed networks by incorporating the objectives of external tasks into the learning process.Dissertation/Thesis
Doctoral Dissertation Computer Science 2019
43
Hendricks, Dieter. "An online adaptive learning algorithm for optimal trade execution in high-frequency markets." Thesis, 2016. http://hdl.handle.net/10539/21710.
Full textAbstract:
A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy
in the Faculty of Science, School of Computer Science and Applied Mathematics
University of the Witwatersrand. October 2016.Automated algorithmic trade execution is a central problem in modern financial markets, however finding and navigating optimal trajectories in this system is a non-trivial task. Many authors have developed exact analytical solutions by making simplifying assumptions regarding governing dynamics, however for practical feasibility and robustness, a more dynamic approach is needed to capture the spatial and temporal system complexity and adapt as intraday regimes change. This thesis aims to consolidate four key ideas: 1) the financial market as a complex adaptive system, where purposeful agents with varying system visibility collectively and simultaneously create and perceive their environment as they interact with it; 2) spin glass models as a tractable formalism to model phenomena in this complex system; 3) the multivariate Hawkes process as a candidate governing process for limit order book events; and 4) reinforcement learning as a framework for online, adaptive learning. Combined with the data and computational challenges of developing an efficient, machine-scale trading algorithm, we present a feasible scheme which systematically encodes these ideas. We first determine the efficacy of the proposed learning framework, under the conjecture of approximate Markovian dynamics in the equity market. We find that a simple lookup table Q-learning algorithm, with discrete state attributes and discrete actions, is able to improve post-trade implementation shortfall by adapting a typical static arrival-price volume trajectory with respect to prevailing market microstructure features streaming from the limit order book. To enumerate a scale-specific state space whilst avoiding the curse of dimensionality, we propose a novel approach to detect the intraday temporal financial market state at each decision point in the Q-learning algorithm, inspired by the complex adaptive system paradigm. A physical analogy to the ferromagnetic Potts model at thermal equilibrium is used to develop a high-speed maximum likelihood clustering algorithm, appropriate for measuring critical or near-critical temporal states in the financial system. State features are studied to extract time-scale-specific state signature vectors, which serve as low-dimensional state descriptors and enable online state detection. To assess the impact of agent interactions on the system, a multivariate Hawkes process is used to measure the resiliency of the limit order book with respect to liquidity-demand events of varying size. By studying the branching ratios associated with key quote replenishment intensities following trades, we ensure that the limit order book is expected to be resilient with respect to the maximum permissible trade executed by the agent. Finally we present a feasible scheme for unsupervised state discovery, state detection and online learning for high-frequency quantitative trading agents faced with a multifeatured, asynchronous market data feed. We provide a technique for enumerating the state space at the scale at which the agent interacts with the system, incorporating the effects of a live trading agent on limit order book dynamics into the market data feed, and hence the perceived state evolution.
LG2017
44
Valko, Michal. "Adaptive Graph-Based Algorithms for Conditional Anomaly Detection and Semi-Supervised Learning." Phd thesis, 2011. http://tel.archives-ouvertes.fr/tel-00643508.
Full textAbstract:
We develop graph-based methods for semi-supervised learning based on label propagation on a data similarity graph. When data is abundant or arrive in a stream, the problems of computation and data storage arise for any graph-based method. We propose a fast approximate online algorithm that solves for the harmonic solution on an approximate graph. We show, both empirically and theoretically, that good behavior can be achieved by collapsing nearby points into a set of local representative points that minimize distortion. Moreover, we regularize the harmonic solution to achieve better stability properties. We also present graph-based methods for detecting conditional anomalies and apply them to the identification of unusual clinical actions in hospitals. Our hypothesis is that patient-management actions that are unusual with respect to the past patients may be due to errors and that it is worthwhile to raise an alert if such a condition is encountered. Conditional anomaly detection extends standard unconditional anomaly framework but also faces new problems known as fringe and isolated points. We devise novel nonparametric graph-based methods to tackle these problems. Our methods rely on graph connectivity analysis and soft harmonic solution. Finally, we conduct an extensive human evaluation study of our conditional anomaly methods by 15 experts in critical care.
45
Liang, Yen-Lun, and 梁晏綸. "Empirical studies on the online learning algorithms based on combining weight noise injection and weight decay." Thesis, 2010. http://ndltd.ncl.edu.tw/handle/81560444116082287065.
Full textAbstract:
碩士國立中興大學
科技管理研究所
98
While injecting weight noise during training have been widely adopted in attaining fault tolerant neural newtorks, theoretical and empirical studies on the online algorithms developed based on these strategies have yet to be complete. In this thesis, we will investigate two important aspects in regard to the online learning algorithms based on combining weight noise injection and weight decay. Multiplicative weight noise and additive weight noise are considered seperately. The convergence behaviors and the performance of those learning algorithms are investigated via intensive computer simulations. It is found that (i) the online learning algorithm based on purely multiplicative weight noise injection does not converge, (ii) the algorithms combining weight noise injection and weight decay exhibit better convergence behaviors than their pure weight noise injection counterparts, and (iii) the neural networks attained by these algorithms combining weight noise injection and weight decay showing better fault tolerance abilities than the neural networks attained by the pure weight noise injection-based algorithms. The contributions of these results are two folds. First, part of these empirical results complement the recent findings from Ho, Leung & Sum on the convergence behaviors of the weight noise injection-based learning algorithms. Second, another part of the results which is in regard to the fault tolerance ability are new in the area. Finally, one should note that the results presented in this thesis also bring out an important message adding weight decay during training. Weight decay is not just can improve the convergence of an algorithm, but also can improve the weight noise tolerance ability of a neural network that is attained by these online algorithms.
46
Hsiao, Yi-Cheng, and 蕭義橙. "A statistical document classification system based on machine learning algorithms: Architecture and application in Facebook online discussion group." Thesis, 2017. http://ndltd.ncl.edu.tw/handle/57322n.
Full textAbstract:
碩士國立交通大學
教育研究所
106
The aim of this study is to develop a Chinese document classification systems for judging whether the content of the text is statistically relevant by means of machine learning algorithms. And the system is applied to the Facebook online discussion group in statistics course, classify posts and comments in the group is statistically relevant or not. Finally, this study will compare the reliability between machine classification and manual classification to explore whether the machine can achieve similar classification with humans. The experimental results show that the accuracy of the machine classification model is between .917 and .950, and the reliability of machine classification and manual classification is between .522 and .760, which means that the machine has high classification accuracy and have the potential to replace manual classification.
47
Chen, Jun-Hong, and 陳俊宏. "On the Prediction of Financial Time Series via Online Machine Learning Algorithms — An Example of S&P 500 Index Component Stocks." Thesis, 2016. http://ndltd.ncl.edu.tw/handle/dacay2.
Full textAbstract:
碩士臺北市立大學
資訊科學系碩士在職專班
104
This thesis studies online machine learning algorithms to the prediction problem for financial time series. Much research has been conducted in the field of time series analysis, in which many machine learning algorithms have been adopted to predict the trend of time series. However, as time goes by and more data become available, traditional machine learning algorithms need to merge the old data and the new one to a new training data, and re-train the model on the new training data. An online algorithm is one that can process its input piece-by-piece in a serial fashion, which makes it suitable for time series analysis. Therefore, in this study, we attempt to use several online machine learning algorithms to analyze the trend of S&P 500 index component stock prices. Experimental results show that there are small differences in terms of the accuracy between the offline and online algorithms; furthermore, the training time for the online learning algorithms is much faster than that for the offline algorithms, as the training data increases with time.
48
Singh, Ravinder. "Extracting Human Behaviour and Personality Traits from Social Media." Thesis, 2021. https://vuir.vu.edu.au/42639/.
Full textAbstract:
Online social media has evolved as an integral part of human society. It facilitates collaboration and information flow, and has emerged as a crucial instrument for business and government organizations alike. Online platforms are being used extensively for work, entertainment, collaboration and communication. These positive aspects are, however, overshadowed by their shortcomings. With the constant evolution and expansion of social media platforms, a significant shift has occurred in the way some humans interact with others. Online social media platforms have inadvertently emerged as networking hubs for individuals exhibiting antisocial behaviour (ASB), putting vulnerable groups of people at risk. Online ASB is one of the most common personality disorders seen on these platforms, and is challenging to address due to its complexities.
Human rights are the keystones of sturdy communities. Respect for these rights, based on the values of equality, dignity and appreciation, is vital and an integral part of strong societies. Every individual has a fundamental right to freely participate in all legal activities, including socializing in both the physical and online worlds. ASB, ranging from threatening, aggression, disregard for safety and failure to conform to lawful behaviour, deter such participation and must be dealt with accordingly. Online ASB is the manifestation of everyday sadism and violates the elementary rights (to which all individuals are entitled) of Its victims. Not only does it interfere with social participation, it also forces individuals into anxiety, depression and suicidal ideation. The consequences of online ASB for victims' and families' mental health are often far-reaching, severe and long-lasting, and can even create a social welfare burden. The behaviour can, not only inhibit constructive user participation with social media, it defies the sole purpose of these platforms: to facilitate communication and collaboration at scale. ASB needs to be detected and curtailed, encouraging fair user participation and preventing vulnerable groups of people from falling victim to such behaviour.
Considering the large variety, high contribution speed and high volume of social media data, a manual approach to detecting and classifying online ASB is not a feasible option. Furthermore, a traditional approach based on a pre-defined lexicon and rule-based feature engineering may still fall short of capturing the subtle and latent features of the diverse and enormous volume of social media data. State-of-the-art deep learning, which is a sub-field of machine learning, has produced astonishing results in numerous text classification undertakings, and has outperformed the aforementioned techniques. However, given the complexity associated with implementing deep learning algorithms and their relatively recent development, models based on the technology have significantly been under-utilized when working with online behaviour studies. Specifically, no prior study has undertaken the task of fine-grained and user- generated social media content classification related to online ASB utilizing the deep learning technology.
This thesis introduces a novel three-part framework, based on deep learning, with the objectives of: i) Detecting behaviour and personality traits from online platforms; (ii) Binary detection of online antisocial behaviour and (iii) Multiclass antisocial behaviour detection from social media corpora. A high accuracy classification model is presented proceeded by extensive experimentation with different machine learning and deep learning algorithms, fine tuning of hyper- parameters, and using different feature extraction techniques. Disparate behaviour and personality traits, including ASB and its four variants are detected with a significantly high accuracy from online social media platforms. Along the way, three medium-sized gold standard benchmark data set have been constructed. The proposed approach is seminal and offers a step towards efficient and effective methods of online ASB prevention. The approach and the findings within this thesis are significant and crucial as these lay the groundwork for detecting and eliminating all types of undesirable and unacceptable social behaviour traits from online platforms.
49
Azami, Sajjad. "Exploring fair machine learning in sequential prediction and supervised learning." Thesis, 2020. http://hdl.handle.net/1828/12098.
Full textAbstract:
Algorithms that are being used in sensitive contexts such as deciding to give a job offer or giving inmates parole should be accurate as well as being non-discriminatory. The latter is important especially due to emerging concerns about automatic decision making being unfair to individuals belonging to certain groups. The machine learning literature has seen a rapid evolution in research on this topic. In this thesis, we study various problems in sequential decision making motivated by challenges in algorithmic fairness. As part of this thesis, we modify the fundamental framework of prediction with expert advice. We assume a learning agent is making decisions using the advice provided by a set of experts while this set can shrink. In other words, experts can become unavailable due to scenarios such as emerging anti-discriminatory laws prohibiting the learner from using experts detected to be unfair. We provide efficient algorithms for this setup, as well as a detailed analysis of the optimality of them. Later we explore a problem concerned with providing any-time fairness guarantees using the well-known exponential weights algorithm, which leads to an open question about a lower bound on the cumulative loss of exponential weights algorithm. Finally, we introduce a novel fairness notion for supervised learning tasks motivated by the concept of envy-freeness. We show how this notion might bypass certain issues of existing fairness notions such as equalized odds. We provide solutions for a simplified version of this problem and insights to deal with further challenges that arise by adopting this notion.Graduate
50
Turgeon, Stéphanie. "L’analyse appliquée du comportement en autisme et ses enjeux : une évaluation du potentiel de la technologie pour améliorer la pratique et la recherche." Thesis, 2021. http://hdl.handle.net/1866/25604.
Full textAbstract:
Le trouble du spectre de l’autisme (TSA) est un trouble neurodéveloppemental caractérisé par des déficits importants de la communication sociale et des interactions sociales ainsi que par la présence de comportements ou d'intérêts restreints et répétitifs. Les données empiriques suggèrent que les interventions découlant de l’analyse appliquée du comportement (AAC) sont les plus efficaces pour intervenir auprès des personnes ayant un TSA. Néanmoins, certaines lacunes en lien avec les interventions découlant de l’analyse du comportement existent. Notamment, le manque d’accessibilité aux services, le manque de connaissances quant aux facteurs sous-jacents à l’efficacité des interventions et les perceptions divergentes de l’AAC freinent son adoption à plus grande échelle. Cette thèse comprend trois études qui mettent à profit la technologie pour mieux comprendre ou améliorer ces enjeux entourant l’AAC.
Dans le cadre ma première étude, les effets d’une formation interactive en ligne qui vise à enseigner aux parents des stratégies découlant de l’AAC pour réduire les comportements problématiques de leur enfant ont été évalués à l’aide d’un devis randomisé contrôlé avec liste d’attente. Les résultats de cette étude soutiennent le potentiel et l’efficacité de la formation pour augmenter la fréquence d’utilisation de stratégies d’intervention découlant de l’AAC par les parents ainsi que pour réduire l’occurrence et la sévérité des comportements problématiques de leur enfant. En revanche, aucune différence significative n’a été observée pour la mesure des pratiques parentales. Certains enjeux éthiques et pratiques entourant la dissémination de la formation en ligne complètement auto-guidées sont discutés.
La deuxième étude de ma thèse doctorale visait donc à montrer comment utiliser des algorithmes d’apprentissage automatique pour identifier les personnes qui sont plus enclines à observer des améliorations suivant une intervention. Plus spécifiquement, l’utilisation de quatre algorithmes d’apprentissage automatique pour prédire les participants ayant pris part à la première étude de cette thèse qui étaient les plus propices à rapporter une diminution des comportements problématiques de leur enfant est démontrée. Cette étude soutient que des algorithmes d’apprentissage automatique peuvent être utilisés avec de petits échantillons pour soutenir la prise de décision des cliniciens et des chercheurs.
La troisième étude cette thèse visait à quantifier l’information sur l’AAC publiée dans quatre sous-forums d’un forum internet, une ressource en ligne souvent utilisée par les familles pour identifier des interventions à utiliser après de leur enfant. Pour atteindre cet objectif, une procédure de forage de données a été réalisée. Les analyses de cette étude appuient que les parents qui fréquentent le forum sont exposés à une proportion importante de messages présentant une désapprobation de l’AAC pour intervenir auprès des personnes ayant un TSA ou bien une description inexacte des principes, méthodes, procédures ou interventions qui en découlent.
Ensemble, les études effectuées dans le cadre de ma thèse doctorale mettent en évidence les bienfaits de la technologie pour l’intervention psychosociale, tant au niveau de l’évaluation que de l’intervention et du transfert de connaissances. Comme souligné dans les trois études de cette thèse, chacun des outils utilisés présente des limites et doit donc être utilisé pour soutenir les cliniciens et les chercheurs, et non pour remplacer leurs interventions et leur jugement clinique. Les études futures doivent continuer à s’intéresser à l’efficacité des outils technologiques, mais également aux facteurs sous-jacents qui favoriseront leur utilisation et aux considérations éthiques liées à leur emploi.Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by significant deficits in social communication and social interactions and by the presence of restricted and repetitive behaviors or interests. Empirical evidence suggests that interventions based on applied behavior analysis (ABA) are the most effective for treating individuals with ASD. Nevertheless, interventions based on behavior analysis present some issues. In particular, intervention services are hard to access, knowledge about the underlying factors of the effectiveness of interventions is lacking and divergent perceptions about of ABA hamper the adoption of the science. This dissertation includes three studies in which technology is used to better understand or improve these issues regarding ABA. As part of my first study, the effects of a fully self-guided interactive web training (IWT) developed for teaching parents of children with ASD ABA-derived strategies to reduce their child's challenging behaviors were evaluated using a randomized waitlist trial. The results of this study support the effectiveness of the IWT for increasing the frequency of parents’ use of behavioral interventions as well as for reducing the frequency and severity of their child’s challenging behaviors. In contrast, no significant difference was observed for the measurement of parenting practices. Ethical and practical consideration regarding the dissemination of fully self-guided online trainings are discussed. The second study of my doctoral thesis aimed to show how to use machine learning algorithms to predict individuals who were most likely to improve following an intervention. Specifically, a demonstration of how to implement four machine learning algorithms to predict the participants from my first study who were the most likely to report a decrease in their child's iv challenging behaviors. This study argues that machine learning algorithms can be used with small samples to support clinicians’ and researchers’ decision making. The third study of my dissertation aimed to quantify the information about ABA published on four subforums of an internet forum; an online resource often used by families to identify potential interventions for their child. This goal was achieved through the use of a data mining procedure. The analyses showed that parents who visited the forum were exposed to a significant proportion of messages that disapproved of ABA for individuals with ASD or that inaccurately described its underlying principles, methods, procedures, or interventions. Together, the studies carried out as part of my doctoral dissertation highlight the benefits of technology to support assessments, interventions, and knowledge gains or transfer within psychosocial practices. As highlighted in the three studies of this dissertation, each of the tools used presents limitations and should therefore be used to support clinicians and researchers, and should not replace their interventions and clinical judgment. Future studies should continue to focus on the effectiveness of technological tools and on the underlying factors that will promote their use. Finally, researchers must reflect on the ethical considerations related to use of technology when working with humans.