Siga este enlace para ver otros tipos de publicaciones sobre el tema: Learning dynamical systems.
Tesis sobre el tema "Learning dynamical systems"
Crea una cita precisa en los estilos APA, MLA, Chicago, Harvard y otros
Consulte los 50 mejores tesis para su investigación sobre el tema "Learning dynamical systems".
Junto a cada fuente en la lista de referencias hay un botón "Agregar a la bibliografía". Pulsa este botón, y generaremos automáticamente la referencia bibliográfica para la obra elegida en el estilo de cita que necesites: APA, MLA, Harvard, Vancouver, Chicago, etc.
También puede descargar el texto completo de la publicación académica en formato pdf y leer en línea su resumen siempre que esté disponible en los metadatos.
Explore tesis sobre una amplia variedad de disciplinas y organice su bibliografía correctamente.
1
Preen, Richard John. "Dynamical genetic programming in learning classifier systems". Thesis, University of the West of England, Bristol, 2011. http://eprints.uwe.ac.uk/25852/.
Texto completoResumen
Learning Classifier Systems (LCS) traditionally use a ternary encoding to generalise over the environmental inputs and to associate appropriate actions. However, a number of schemes have been presented beyond this, ranging from integers to artificial neural networks. This thesis investigates the use of Dynamical Genetic Programming (DGP) as a knowledge representation within LCS. DGP is a temporally dynamic, graph-based, symbolic representation. Temporal dynamism has been identified as an important aspect in biological systems, artificial life, and cognition in general. Furthermore, discrete dynamical systems have been found to exhibit inherent content-addressable memory. In this thesis, the collective emergent behaviour of ensembles of such dynamical function networks are herein shown to be exploitable toward solving various computational tasks. Significantly, it is shown possible to exploit the variable-length, adaptive memory existing inherently within the networks under an asynchronous scheme, and where all new parameters introduced are self-adaptive. It is shown possible to exploit the collective mechanics to solve both discrete and continuous-valued reinforcement learning problems, and to perform symbolic regression. In particular, the representation is shown to provide improved performance beyond a traditional Genetic Programming benchmark on a number of a composite polynomial regression tasks. Superior performance to previously published techniques is also shown in a continuous-input-output reinforcement learning problem. Finally, it is shown possible to perform multi-step-ahead predictions of a financial time-series by repeatedly sampling the network states at succeeding temporal intervals.
2
Ferizbegovic, Mina. "Robust learning and control of linear dynamical systems". Licentiate thesis, KTH, Reglerteknik, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-280121.
Texto completoResumen
We consider the linear quadratic regulation problem when the plant is an unknown linear dynamical system. We present robust model-based methods based on convex optimization, which minimize the worst-case cost with respect to uncertainty around model estimates. To quantify uncertainty, we derive a methodbased on Bayesian inference, which is directly applicable to robust control synthesis.We focus on control policies that can be iteratively updated after sequentially collecting data. More specifically, we seek to design control policies that balance exploration (reducing model uncertainty) and exploitation (control of the system) when exploration must be safe (robust).First, we derive a robust controller to minimize the worst-case cost, with high probability, given the empirical observation of the system. This robust controller synthesis is then used to derive a robust dual controller, which updates its control policy after collecting data. An episode in which data is collected is called exploration, and the episode using an updated control policy is exploitation. The objective is to minimize the worst-case cost of the updated control policy, requiring that a given exploration budget constrains the worst-case cost during exploration.We look into robust dual control in both finite and infinite horizon settings. The main difference between the finite and infinite horizon settings is that the latter does not consider the length of the exploration and exploitation phase, but it rather approximates the cost using the infinite horizon cost. In the finite horizon setting, we discuss how different exploration lengths affect the trade-off between exploration and exploitation.Additionally, we derive methods that balance exploration and exploitation to minimize the cumulative worst-case cost for a fixed number of episodes. In this thesis, we refer to such a problem as robust reinforcement learning. Essentially, it is a robust dual controller aiming to minimize the cumulative worst-case cost, and that updates its control policy in each episode.Numerical experiments show that the proposed methods have better performance compared to existing state-of-the-art algorithms. Moreover, experiments also indicate that the exploration prioritizes the uncertainty reduction in the parameters that matter most for control.QC 20200904
3
Mazzoleni, Mirko (ORCID:0000-0002-7116-135X). "Learning meets control. Data analytics for dynamical systems". Doctoral thesis, Università degli studi di Bergamo, 2018. http://hdl.handle.net/10446/104812.
Texto completoResumen
System identification has always been one of the main research focuses of the control community, since the early steps of the automatic control field.
The development of a dynamical system’s models from experimental data is instrumental for understanding the plant under study and designing its model-based control scheme. In the last decade, a cross-fertilization began between the System Identification and the Statistical Learning communities.
This led firstly to the introduction of regularization techniques in system identification, and, more recently, to the application of kernel methods to dynamical system learning. This thesis further investigates the roles that learning methods can have in the control science. In the first part, we lay the theoretical foundations of a new kernel-based regularization method for Nonlinear Finite Impulse Response (NFIR) system identification. The method, called Semi-Supervised Identification (SSI), relies on the manifold spanned by the system’s inputs. This manifold is constructed by using not only the measured input/output data, but also inputs data for which there is no corresponding outputs. The effect of this rationale is to impose prior information on the system structure, in the form of local smoothness assumptions. This differs from standard Tikhonov regularization, which imposes a global smoothness behaviour on the learned function. The second part of this work presents practical applications of how statistical learning methods can be used to face control and estimation problems. The case studies span a variety of different applications, from fault detection of electro-mechanical actuators, to clustering methodologies and pure forecasting challanges.
4
Izquierdo, Eduardo J. "The dynamics of learning behaviour : a situated, embodied, and dynamical systems approach". Thesis, University of Sussex, 2008. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.488595.
Texto completo
5
Mussmann, Thomas Frederick. "Data Driven Learning of Dynamical Systems Using Neural Networks". The Ohio State University, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=osu1618589877977348.
Texto completo
6
Lindsten, Fredrik. "Particle filters and Markov chains for learning of dynamical systems". Doctoral thesis, Linköpings universitet, Reglerteknik, 2013. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-97692.
Texto completoResumen
Sequential Monte Carlo (SMC) and Markov chain Monte Carlo (MCMC) methods provide computational tools for systematic inference and learning in complex dynamical systems, such as nonlinear and non-Gaussian state-space models. This thesis builds upon several methodological advances within these classes of Monte Carlo methods.Particular emphasis is placed on the combination of SMC and MCMC in so called particle MCMC algorithms. These algorithms rely on SMC for generating samples from the often highly autocorrelated state-trajectory. A specific particle MCMC algorithm, referred to as particle Gibbs with ancestor sampling (PGAS), is suggested. By making use of backward sampling ideas, albeit implemented in a forward-only fashion, PGAS enjoys good mixing even when using seemingly few particles in the underlying SMC sampler. This results in a computationally competitive particle MCMC algorithm. As illustrated in this thesis, PGAS is a useful tool for both Bayesian and frequentistic parameter inference as well as for state smoothing. The PGAS sampler is successfully applied to the classical problem of Wiener system identification, and it is also used for inference in the challenging class of non-Markovian latent variable models.Many nonlinear models encountered in practice contain some tractable substructure. As a second problem considered in this thesis, we develop Monte Carlo methods capable of exploiting such substructures to obtain more accurate estimators than what is provided otherwise. For the filtering problem, this can be done by using the well known Rao-Blackwellized particle filter (RBPF). The RBPF is analysed in terms of asymptotic variance, resulting in an expression for the performance gain offered by Rao-Blackwellization. Furthermore, a Rao-Blackwellized particle smoother is derived, capable of addressing the smoothing problem in so called mixed linear/nonlinear state-space models. The idea of Rao-Blackwellization is also used to develop an online algorithm for Bayesian parameter inference in nonlinear state-space models with affine parameter dependencies.CNDM
CADICS
7
Mao, Weize. "DATA-DRIVEN LEARNING OF UNKNOWN DYNAMICAL SYSTEMS WITH MISSING INFORMATION". The Ohio State University, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=osu1619097149112362.
Texto completo
8
Passey, Jr David Joseph. "Growing Complex Networks for Better Learning of Chaotic Dynamical Systems". BYU ScholarsArchive, 2020. https://scholarsarchive.byu.edu/etd/8146.
Texto completoResumen
This thesis advances the theory of network specialization by characterizing the effect of network specialization on the eigenvectors of a network. We prove and provide explicit formulas for the eigenvectors of specialized graphs based on the eigenvectors of their parent graphs. The second portion of this thesis applies network specialization to learning problems. Our work focuses on training reservoir computers to mimic the Lorentz equations. We experiment with random graph, preferential attachment and small world topologies and demonstrate that the random removal of directed edges increases predictive capability of a reservoir topology. We then create a new network model by growing networks via targeted application of the specialization model. This is accomplished iteratively by selecting top preforming nodes within the reservoir computer and specializing them. Our generated topology out-preforms all other topologies on average.
9
Bézenac, Emmanuel de. "Modeling physical processes with deep learning : a dynamical systems approach". Electronic Thesis or Diss., Sorbonne université, 2021. http://www.theses.fr/2021SORUS203.
Texto completoResumen
L'apprentissage profond s'impose comme un outil prédominant pour l'IA, avec de nombreuses applications fructueuses pour des taches où les données sont abondantes et l'accès aux connaissances préalables est difficile. Cependant ce n'est pas encore le cas dans le domaine des sciences naturelles, et encore moins pour l'étude des systèmes dynamiques. En effet, ceux-ci font l'objet d'études depuis des siècles, une quantité considérable de connaissances a ainsi été acquise, et des algorithmes et des méthodes ingénieux ont été développés. Cette thèse a donc deux objectifs principaux. Le premier concerne l'étude du rôle que l'apprentissage profond doit jouer dans ce vaste écosystème de connaissances, de théories et d'outils. Nous tenterons de répondre à cette question générale à travers un problème concret: la modélisation de processus physiques complexes à l'aide de l'apprentissage profond. Le deuxième objectif est en quelque sorte son contraire; il concerne l'analyse les algorithmes d'apprentissage profond à travers le prisme des systèmes dynamiques et des processus physiques, dans le but d'acquérir une meilleure compréhension et de développer de nouveaux algorithmes pour ce domaineDeep Learning has emerged as a predominant tool for AI, and has already abundant applications in fields where data is abundant and access to prior knowledge is difficult. This is not necessarily the case for natural sciences, and in particular, for physical processes. Indeed, these have been the object of study since centuries, a vast amount of knowledge has been acquired, and elaborate algorithms and methods have been developped. Thus, this thesis has two main objectives. The first considers the study of the role that deep learning has to play in this vast ecosystem of knowledge, theory and tools. We will attempt to answer this general question through a concrete problem: the one of modelling complex physical processes, leveraging deep learning methods in order to make up for lacking prior knowledge. The second objective is somewhat its converse: it focuses on how perspectives, insights and tools from the field of study of physical processes and dynamical systems can be applied in the context of deep learning, in order to gain a better understanding and develop novel algorithms
10
Appeltant, Lennert. "Reservoir computing based on delay-dynamical systems". Doctoral thesis, Universitat de les Illes Balears, 2012. http://hdl.handle.net/10803/84144.
Texto completoResumen
Today, except for mathematical operations, our brain functions much faster and more efficient than any supercomputer. It is precisely this form of information processing in neural networks that inspires researchers to create systems that mimic the brain’s information processing capabilities. In this thesis we propose a novel approach to implement these alternative computer architectures, based on delayed feedback. We show that one single nonlinear node with delayed feedback can replace a large network of nonlinear nodes.
First we numerically investigate the architecture and performance of delayed feedback systems as information processing units. Then we elaborate on electronic and opto-electronic implementations of the concept. Next to evaluating their performance for standard benchmarks, we also study task independent properties of the system, extracting information on how to further improve the initial scheme. Finally, some simple modifications are suggested, yielding improvements in terms of speed or performance.
11
Dernsjö, Axel y Wahlström Max Berg. "Data-Driven Learning for Approximating Dynamical Systems Using Deep Neural Networks". Thesis, KTH, Skolan för teknikvetenskap (SCI), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-297685.
Texto completoResumen
In this thesis, a one-step approximation method has been used to produce approximations of two dynamical systems. The two systems considered are a pendulum and a damped dual-mass-spring system. Using a method for a one-step approximation proposed by [15] it is first shown that the state variables of a general dynamical system one time-step ahead can be expressed using a concept called effective increment. The state of the system one time-step ahead then only depends on the previous state and the effective increment, and this effective increment in turn only depends on the previous state and the governing equation of the dynamical system. By introducing the concept of neural networks and surrounding concepts it is presented that a neural network could be trained to approximate this effective increment, thereby negating the need to have a known governing equation when determining the system state. The solution to a general dynamical system can then be approximated using only the trained neural network operator and a state variable to produce the state variable one discrete time-step ahead. When training the neural network operator to approximate the effective increment, the analytical solutions to two dynamical systems are used to produce large amounts of training data on which the network can be trained. Using the optimizer algorithm Adam [8] and the collected training data the network parameters were changed to make the difference between the output of the network and some target value small, the target value, in this case, being the correct state variable one time-step ahead. The results show that training a neural network to be able to produce approximations of a dynamical system is possible, but if one wants to produce more accurate approximations of more complex systems than the ones considered in this thesis, greater care has to be taken when choosing parameters of the network as well as tweaking the hyper-parameters of the optimizer Adam. Furthermore, the structure of the network could be tweaked by changing the number of hidden layers and the number of nodes in them.
12
Yin, Yuan. "Physics-Aware Deep Learning and Dynamical Systems : Hybrid Modeling and Generalization". Electronic Thesis or Diss., Sorbonne université, 2023. http://www.theses.fr/2023SORUS161.
Texto completoResumen
L'apprentissage profond a fait des progrès dans divers domaines et est devenu un outil prometteur pour modéliser les phénomènes dynamiques physiques présentant des relations hautement non linéaires. Cependant, les approches existantes sont limitées dans leur capacité à faire des prédictions physiquement fiables en raison du manque de connaissances préalables et à gérer les scénarios du monde réel où les données proviennent de dynamiques multiples ou sont irrégulièrement distribuées dans le temps et l'espace. Cette thèse vise à surmonter ces limitations dans les directions suivantes: améliorer la modélisation de la dynamique basée sur les réseaux neuronaux en exploitant des modèles physiques grâce à la modélisation hybride ; étendre le pouvoir de généralisation des modèles de dynamique en apprenant les similitudes à partir de données de différentes dynamiques pour extrapoler vers des systèmes invisibles ; et gérer les données de forme libre et prédire continuellement les phénomènes dans le temps et l'espace grâce à la modélisation continue. Nous soulignons la polyvalence des techniques d'apprentissage profond, et les directions proposées montrent des promesses pour améliorer leur précision et leur puissance de généralisation, ouvrant la voie à des recherches futures dans de nouvelles applicationsDeep learning has made significant progress in various fields and has emerged as a promising tool for modeling physical dynamical phenomena that exhibit highly nonlinear relationships. However, existing approaches are limited in their ability to make physically sound predictions due to the lack of prior knowledge and to handle real-world scenarios where data comes from multiple dynamics or is irregularly distributed in time and space. This thesis aims to overcome these limitations in the following directions: improving neural network-based dynamics modeling by leveraging physical models through hybrid modeling; extending the generalization power of dynamics models by learning commonalities from data of different dynamics to extrapolate to unseen systems; and handling free-form data and continuously predicting phenomena in time and space through continuous modeling. We highlight the versatility of deep learning techniques, and the proposed directions show promise for improving their accuracy and generalization power, paving the way for future research in new applications
13
Liu, Xinhe. "Implementation of dynamical systems with plastic self-organising velocity fields". Thesis, Loughborough University, 2015. https://dspace.lboro.ac.uk/2134/19550.
Texto completoResumen
To describe learning, as an alternative to a neural network recently dynamical systems were introduced whose vector fields were plastic and self-organising. Such a system automatically modifies its velocity vector field in response to the external stimuli. In the simplest case under certain conditions its vector field develops into a gradient of a multi-dimensional probability density distribution of the stimuli. We illustrate with examples how such a system carries out categorisation, pattern recognition, memorisation and forgetting without any supervision.
14
Abramova, Ekaterina. "Combining reinforcement learning and optimal control for the control of nonlinear dynamical systems". Thesis, Imperial College London, 2015. http://hdl.handle.net/10044/1/39968.
Texto completoResumen
This thesis presents a novel hierarchical learning framework, Reinforcement Learning Optimal Control, for controlling nonlinear dynamical systems with continuous states and actions. The adapted approach mimics the neural computations that allow our brain to bridge across the divide between symbolic action-selection and low-level actuation control by operating at two levels of abstraction. First, current findings demonstrate that at the level of limb coordination human behaviour is explained by linear optimal feedback control theory, where cost functions match energy and timing constraints of tasks. Second, humans learn cognitive tasks involving learning symbolic level action selection, in terms of both model-free and model-based reinforcement learning algorithms. We postulate that the ease with which humans learn complex nonlinear tasks arises from combining these two levels of abstraction. The Reinforcement Learning Optimal Control framework learns the local task dynamics from naive experience using an expectation maximization algorithm for estimation of linear dynamical systems and forms locally optimal Linear Quadratic Regulators, producing continuous low-level control. A high-level reinforcement learning agent uses these available controllers as actions and learns how to combine them in state space, while maximizing a long term reward. The optimal control costs form training signals for high-level symbolic learner. The algorithm demonstrates that a small number of locally optimal linear controllers can be combined in a smart way to solve global nonlinear control problems and forms a proof-of-principle to how the brain may bridge the divide between low-level continuous control and high-level symbolic action selection. It competes in terms of computational cost and solution quality with state-of-the-art control, which is illustrated with solutions to benchmark problems.
15
Ncube, Israel. "Stochastic approximation of artificial neural network-type learning algorithms, a dynamical systems approach". Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2001. http://www.collectionscanada.ca/obj/s4/f2/dsk3/ftp04/NQ60559.pdf.
Texto completo
16
Chaabene, Walid. "Scalable Structure Learning of Graphical Models". Thesis, Virginia Tech, 2017. http://hdl.handle.net/10919/86263.
Texto completoResumen
Hypothesis-free learning is increasingly popular given the large amounts of data becoming available. Structure learning, a hypothesis-free approach, of graphical models is a field of growing interest due to the power of such models and lack of domain knowledge when applied on complex real-world data. State-of-the-art techniques improve on scalability of structure learning, which is often characterized by a large problem space. Nonetheless, these techniques still suffer computational bottlenecks that are yet to be approached.
In this work, we focus on two popular models: dynamical linear systems and Markov random fields. For each case, we investigate major computational bottlenecks of baseline learning techniques. Next, we propose two frameworks that provide higher scalability using appropriate problem reformulation and efficient structure based heuristics. We perform experiments on synthetic and real data to validate our theoretical analysis. Current results show that we obtain a quality similar to expensive baseline techniques but with higher scalability.Master of Science
17
Wang, Peng. "STOCHASTIC MODELING AND UNCERTAINTY EVALUATION FOR PERFORMANCE PROGNOSIS IN DYNAMICAL SYSTEMS". Case Western Reserve University School of Graduate Studies / OhioLINK, 2017. http://rave.ohiolink.edu/etdc/view?acc_num=case1499788641069811.
Texto completo
18
Marsden, Christopher J. "Nonlinear dynamics of pattern recognition and optimization". Thesis, Loughborough University, 2012. https://dspace.lboro.ac.uk/2134/10694.
Texto completoResumen
We associate learning in living systems with the shaping of the velocity vector field of a dynamical system in response to external, generally random, stimuli. We consider various approaches to implement a system that is able to adapt the whole vector field, rather than just parts of it - a drawback of the most common current learning systems: artificial neural networks. This leads us to propose the mathematical concept of self-shaping dynamical systems. To begin, there is an empty phase space with no attractors, and thus a zero velocity vector field. Upon receiving the random stimulus, the vector field deforms and eventually becomes smooth and deterministic, despite the random nature of the applied force, while the phase space develops various geometrical objects. We consider the simplest of these - gradient self-shaping systems, whose vector field is the gradient of some energy function, which under certain conditions develops into the multi-dimensional probability density distribution of the input. We explain how self-shaping systems are relevant to artificial neural networks. Firstly, we show that they can potentially perform pattern recognition tasks typically implemented by Hopfield neural networks, but without any supervision and on-line, and without developing spurious minima in the phase space. Secondly, they can reconstruct the probability density distribution of input signals, like probabilistic neural networks, but without the need for new training patterns to have to enter the network as new hardware units. We therefore regard self-shaping systems as a generalisation of the neural network concept, achieved by abandoning the "rigid units - flexible couplings'' paradigm and making the vector field fully flexible and amenable to external force. It is not clear how such systems could be implemented in hardware, and so this new concept presents an engineering challenge. It could also become an alternative paradigm for the modelling of both living and learning systems. Mathematically it is interesting to find how a self shaping system could develop non-trivial objects in the phase space such as periodic orbits or chaotic attractors. We investigate how a delayed vector field could form such objects. We show that this method produces chaos in a class systems which have very simple dynamics in the non-delayed case. We also demonstrate the coexistence of bounded and unbounded solutions dependent on the initial conditions and the value of the delay. Finally, we speculate about how such a method could be used in global optimization.
19
Murray, Lawrence. "Bayesian learning of continuous time dynamical systems with applications in functional magnetic resonance imaging". Thesis, University of Edinburgh, 2009. http://hdl.handle.net/1842/4157.
Texto completoResumen
Temporal phenomena in a range of disciplines are more naturally modelled in continuous-time than coerced into a discrete-time formulation. Differential systems form the mainstay of such modelling, in fields from physics to economics, geoscience to neuroscience. While powerful, these are fundamentally limited by their determinism. For the purposes of probabilistic inference, their extension to stochastic differential equations permits a continuous injection of noise and uncertainty into the system, the model, and its observation. This thesis considers Bayesian filtering for state and parameter estimation in general non-linear, non-Gaussian systems using these stochastic differential models. It identifies a number of challenges in this setting over and above those of discrete time, most notably the absence of a closed form transition density. These are addressed via a synergy of diverse work in numerical integration, particle filtering and high performance distributed computing, engineering novel solutions for this class of model. In an area where the default solution is linear discretisation, the first major contribution is the introduction of higher-order numerical schemes, particularly stochastic Runge-Kutta, for more efficient simulation of the system dynamics. Improved runtime performance is demonstrated on a number of problems, and compatibility of these integrators with conventional particle filtering and smoothing schemes discussed. Finding compatibility for the smoothing problem most lacking, the major theoretical contribution of the work is the introduction of two novel particle methods, the kernel forward-backward and kernel two-filter smoothers. By harnessing kernel density approximations in an importance sampling framework, these attain cancellation of the intractable transition density, ensuring applicability in continuous time. The use of kernel estimators is particularly amenable to parallelisation, and provides broader support for smooth densities than a sample-based representation alone, helping alleviate the well known issue of degeneracy in particle smoothers. Implementation of the methods for large-scale problems on high performance computing architectures is provided. Achieving improved temporal and spatial complexity, highly favourable runtime comparisons against conventional techniques are presented. Finally, attention turns to real world problems in the domain of Functional Magnetic Resonance Imaging (fMRI), first constructing a biologically motivated stochastic differential model of the neural and hemodynamic activity underlying the observed signal in fMRI. This model and the methodological advances of the work culminate in application to the deconvolution and effective connectivity problems in this domain.
20
Chakeri, Alireza. "Scalable Unsupervised Learning with Game Theory". Scholar Commons, 2017. http://scholarcommons.usf.edu/etd/6616.
Texto completoResumen
Recently dominant sets, a generalization of the notion of the maximal clique to edge-weighted graphs, have proven to be an effective tool for unsupervised learning and have found applications in different domains. Although, they were initially established using optimization and graph theory concepts, recent work has shown fascinating connections with evolutionary game theory, that leads to the clustering game framework. However, considering size of today's data sets, existing methods need to be modified in order to handle massive data. Hence, in this research work, first we address the limitations of the clustering game framework for large data sets theoretically. We propose a new important question for the clustering community ``How can a cluster of a subset of a dataset be a cluster of the entire dataset?''. We show that, this problem is a coNP-hard problem in a clustering game framework. Thus, we modify the definition of a cluster from a stable concept to a non-stable but optimal one (Nash equilibrium). By experiments we show that this relaxation does not change the qualities of the clusters practically.
Following this alteration and the fact that equilibriums are generally compact subsets of vertices, we design an effective strategy to find equilibriums representing well distributed clusters. After finding such equilibriums, a linear game theoretic relation is proposed to assign vertices to the clusters and partition the graph. However, the method inherits a space complexity issue, that is the similarities between every pair of objects are required which proves practically intractable for large data sets. To overcome this limitation, after establishing necessary theoretical tools for a special type of graphs that we call vertex-repeated graphs, we propose the scalable clustering game framework. This approach divides a data set into disjoint tractable size chunks. Then, the exact clusters of the entire data are approximated by the clusters of the chunks. In fact, the exact equilibriums of the entire graph is approximated by the equilibriums of the subsets of the graph. We show theorems that enable significantly improved time complexity for the model. The applications include, but are not limited to, the maximum weight clique problem, large data clustering and image segmentation. Experiments have been done on random graphs and the DIMACS benchmark for the maximum weight clique problem and on magnetic resonance images (MRI) of the human brain consisting of about 4 million examples for large data clustering. Also, on the Berkeley Segmentation Dataset, the proposed method achieves results comparable to the state of the art, providing a parallel framework for image segmentation and without any training phase. The results show the effectiveness and efficiency of our approach.
In another part of this research work, we generalize the clustering game method to cluster uncertain data where the similarities between the data points are not exactly known, that leads to the uncertain clustering game framework. Here, contrary to the ensemble clustering approaches, where the results of different similarity matrices are combined, we focus on the average utilities of an uncertain game. We show that the game theoretical solutions provide stable clusters even in the presence of severe uncertainties. In addition, based on this framework, we propose a novel concept in uncertain data clustering so that every subset of objects can have a ''cluster degree''. Extensive experiments on real world data sets, as well as on the Berkeley image segmentation dataset, confirm the performance of the proposed method.
And finally, instead of dividing a graph into chunks to make the clustering scalable, we study the effect of the spectral sparsification method based on sampling by effective resistance on the clustering outputs. Through experimental and theoretical observations, we show that the clustering results obtained from sparsified graphs are very similar to the results of the original non-sparsified graphs. The rand index is always at about 0.9 to 0.99 in our experiments even when lots of sparsification is done.
21
Calliess, Jan-Peter. "Conservative decision-making and inference in uncertain dynamical systems". Thesis, University of Oxford, 2014. http://ora.ox.ac.uk/objects/uuid:b7206c3a-8d76-4454-a258-ea1e5bd1c63e.
Texto completoResumen
The demand for automated decision making, learning and inference in uncertain, risk sensitive and dynamically changing situations presents a challenge: to design computational approaches that promise to be widely deployable and flexible to adapt on the one hand, while offering reliable guarantees on safety on the other. The tension between these desiderata has created a gap that, in spite of intensive research and contributions made from a wide range of communities, remains to be filled. This represents an intriguing challenge that provided motivation for much of the work presented in this thesis. With these desiderata in mind, this thesis makes a number of contributions towards the development of algorithms for automated decision-making and inference under uncertainty. To facilitate inference over unobserved effects of actions, we develop machine learning approaches that are suitable for the construction of models over dynamical laws that provide uncertainty bounds around their predictions. As an example application for conservative decision-making, we apply our learning and inference methods to control in uncertain dynamical systems. Owing to the uncertainty bounds, we can derive performance guarantees of the resulting learning-based controllers. Furthermore, our simulations demonstrate that the resulting decision-making algorithms are effective in learning and controlling under uncertain dynamics and can outperform alternative methods. Another set of contributions is made in multi-agent decision-making which we cast in the general framework of optimisation with interaction constraints. The constraints necessitate coordination, for which we develop several methods. As a particularly challenging application domain, our exposition focusses on collision avoidance. Here we consider coordination both in discrete-time and continuous-time dynamical systems. In the continuous-time case, inference is required to ensure that decisions are made that avoid collisions with adjustably high certainty even when computation is inevitably finite. In both discrete-time and finite-time settings, we introduce conservative decision-making. That is, even with finite computation, a coordination outcome is guaranteed to satisfy collision-avoidance constraints with adjustably high confidence relative to the current uncertain model. Our methods are illustrated in simulations in the context of collision avoidance in graphs, multi-commodity flow problems, distributed stochastic model-predictive control, as well as in collision-prediction and avoidance in stochastic differential systems. Finally, we provide an example of how to combine some of our different methods into a multi-agent predictive controller that coordinates learning agents with uncertain beliefs over their dynamics. Utilising the guarantees established for our learning algorithms, the resulting mechanism can provide collision avoidance guarantees relative to the a posteriori epistemic beliefs over the agents' dynamics.
22
Lin, Jing Ph D. Massachusetts Institute of Technology. "Bayesian learning for high-dimensional nonlinear dynamical systems : methodologies, numerics and applications to fluid flows". Thesis, Massachusetts Institute of Technology, 2020. https://hdl.handle.net/1721.1/132760.
Texto completoResumen
Thesis: Ph. D. in Mechanical Engineering and Computation, Massachusetts Institute of Technology, Department of Mechanical Engineering, September, 2020Cataloged from the official PDF of thesis.
Includes bibliographical references (pages 553-567).
The rapidly-growing computational power and the increasing capability of uncertainty quantification, statistical inference, and machine learning have opened up new opportunities for utilizing data to assist, identify and refine physical models. In this thesis, we focus on Bayesian learning for a particular class of models: high-dimensional nonlinear dynamical systems, which have been commonly used to predict a wide range of transient phenomena including fluid flows, heat transfer, biogeochemical dynamics, and other advection-diffusion-reaction-based transport processes. Even though such models often express the differential form of fundamental laws, they commonly contain uncertainty in their initial and boundary values, parameters, forcing and even formulation. Learning such components from sparse observation data by principled Bayesian inference is very challenging due to the systems' high-dimensionality and nonlinearity. We systematically study the theoretical and algorithmic properties of a Bayesian learning methodology built upon previous efforts in our group to address this challenge. Our systematic study breaks down into the three hierarchical components of the Bayesian learning and we develop new numerical schemes for each. The first component is on uncertainty quantification for stochastic dynamical systems and fluid flows. We study dynamic low-rank approximations using the dynamically orthogonal (DO) equations including accuracy and computational costs, and develop new numerical schemes for re-orthonormalization, adaptive subspace augmentation, residual-driven closure, and stochastic Navier-Stokes integration. The second part is on Bayesian data assimilation, where we study the properties of and connections among the different families of nonlinear and non-Gaussian filters. We derive an ensemble square-root filter based on minimal-correction second-moment matching that works especially well under the adversity of small ensemble size, sparse observations and chaotic dynamics. We also obtain a localization technique for filtering with high-dimensional systems that can be applied to nonlinear non-Gaussian inference with both brute force Monte Carlo (MC) and reduced subspace modeling in a unified way. Furthermore, we develop a mutual-information-based adaptive sampling strategy for filtering to identify the most informative observations with respect to the state variables and/or parameters, utilizing the sub-modularity of mutual information due to the conditional independence of observation noise. The third part is on active Bayesian model learning, where we have a discrete set of candidate dynamical models and we infer the model formulation that best explains the data using principled Bayesian learning. To predict the observations that are most useful to learn the model formulation, we further extend the above adaptive sampling strategy to identify the data that are expected to be most informative with respect to both state variables and the uncertain model identity. To investigate and showcase the effectiveness and efficiency of our theoretical and numerical advances for uncertainty quantification, Bayesian data assimilation, and active Bayesian learning with stochastic nonlinear high-dimensional dynamical systems, we apply our dynamic data-driven reduced subspace approach to several dynamical systems and compare our results against those of brute force MC and other existing methods. Specifically, we analyze our advances using several drastically different dynamical regimes modeled by the nonlinear Lorenz-96 ordinary differential equations as well as turbulent bottom gravity current dynamics modeled by the 2-D unsteady incompressible Reynolds-averaged Navier-Stokes (RANS) partial differential equations. We compare the accuracy, efficiency, and robustness of different methodologies and algorithms. With the Lorenz- 96 system, we show how the performance differs under periodic, weakly chaotic, and very chaotic dynamics and under different observation layouts. With the bottom gravity current dynamics, we show how model parameters, domain geometries, initial fields, and boundary forcing formulations can be identified and how the Bayesian methodology performs when the candidate model space does not contain the true model. The results indicate that our active Bayesian learning framework can better infer the state variables and dynamical model identity with fewer observations than many alternative approaches in the literature.
by Jing Lin.
Ph. D. in Mechanical Engineering and Computation
Ph.D.inMechanicalEngineeringandComputation Massachusetts Institute of Technology, Department of Mechanical Engineering
23
Banks, Jess M. "Chaos and Learning in Discrete-Time Neural Networks". Oberlin College Honors Theses / OhioLINK, 2015. http://rave.ohiolink.edu/etdc/view?acc_num=oberlin1445945609.
Texto completo
24
Hefny, Ahmed. "Efficient Methods for Prediction and Control in Partially Observable Environments". Research Showcase @ CMU, 2018. http://repository.cmu.edu/dissertations/1210.
Texto completoResumen
State estimation and tracking (also known as filtering) is an integral part of any system performing inference in a partially observable environment, whether it is a robot that is gauging an environment through noisy sensors or a natural language processing system that is trying to model a sequence of characters without full knowledge of the syntactic or semantic state of the text. In this work, we develop a framework for constructing state estimators. The framework consists of a model class, referred to as predictive state models, and a learning algorithm, referred to as two-stage regression. Our framework is based on two key concepts: (1) predictive state: where our belief about the latent state of the environment is represented as a prediction of future observation features and (2) instrumental regression: where features of previous observations are used to remove sampling noise from future observation statistics, allowing for unbiased estimation of system dynamics. These two concepts allow us to develop efficient and tractable learning methods that reduce the unsupervised problem of learning an environment model to a supervised regression problem: first, a regressor is used to remove noise from future observation statistics. Then another regressor uses the denoised observation features to estimate the dynamics of the environment. We show that our proposed framework enjoys a number of theoretical and practical advantages over existing methods, and we demonstrate its efficacy in a prediction setting, where the task is to predict future observations, as well as a control setting, where the task is to optimize a control policy via reinforcement learning.
25
Grönland, Axel y Möllerstedt Viktor Eriksson. "Robust Reinforcement Learning in Continuous Action/State Space". Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-293879.
Texto completoResumen
In this project we aim to apply Robust Reinforce-ment Learning algorithms, presented by Doya and Morimoto [1],[2], to control problems. Specifically, we train an agent to balancea pendulum in the unstable equilibrium, which is the invertedstate.We investigate the performance of controllers based on twodifferent function approximators. One is quadratic, and the othermakes use of a Radial Basis Function neural network. To achieverobustness we will make use of an approach similar toH∞control, which amounts to introducing an adversary in the controlsystem.By changing the mass of the pendulum after training, we aimedto show as in [2] that the supposedly robust controllers couldhandle this disruption better than its non-robust counterparts.This was not the case. We also added a random disturber signalafter training and performed similar tests, but we were againunable to show robustness.I detta projekt applicerar vi Robust Rein- forcement Learning (RRL) algoritmer, framtagna av Doya och Morimoto [1], [2], på reglerproblem. Målet var att träna en agent att balansera en pendel i det instabila jämviktsläget; det inverterade tillståndet. Vi undersökte prestandan hos regulatorer baserade på två value function approximators. Den ena är kvadratisk och den andra en Radial Basis Function neuralt nätverk. För att skapa robusthet så använder vi en metod som är ekvivalent med H∞ - reglering, som innebär att man introducerar en motståndare i reglersystemet. Genom att ändra pendelns massa efter träning, hoppas vi att som i [2] kunna visa att den förment robusta regulatorn klarar av denna störning bättre än sin icke-robusta mostvarighet. Detta var inte fallet. Vi lade även till en slumpmässig störsignal efter träning och utförde liknande tester, men lyckades inte visa robusthet i detta fall heller.
Kandidatexjobb i elektroteknik 2020, KTH, Stockholm
26
Pagnotta, Murillo. "Living and learning together : integrating developmental systems theory, radical embodied cognitive science, and relational thinking in the study of social learning". Thesis, University of St Andrews, 2018. http://hdl.handle.net/10023/16386.
Texto completoResumen
Behavioural scientists argue that ‘social learning' provides the link between biological phenomena and cultural phenomena because of its role in the ‘cultural transmission' of knowledge among individuals within and across generations. However, leading authors within the social sciences have proposed alternative ways of thinking about social life not founded on the Modern oppositions including nature-culture, biology-culture, body-mind, and individual-society. Similarly, the distinction between a domain of nature and a domain of nurture has also been extensively criticized within biology. Finally, advocates of ‘radical embodied cognitive science' offer an alternative to the representational-computational view of the mind which supports the conventional notion of culture and cultural information. This thesis attempts to integrate developmental systems theory, radical embodied cognitive science, and relational thinking, with the goal to bring the field of social learning closer to these critical theoretical developments. In Chapter 2, I find no justification for the claim that the genome carries information in the sense of specification of biological form. Chapter 3 presents a view of ontogeny as a historical, relational, constructive and contingent process. Chapter 4 uses the notions of environmental information, abilities, affordances, and intentions to make sense of behaviour and learning. In Chapter 5, I argue that the notion of social learning can be understood in terms of relational histories of development rather than in terms of transmission of information. I then report empirical studies investigating behavioural coordination and social learning consistent with this theoretical framework. Chapter 6 presents evidence that dyads in a joint making activity synchronize their attention constrained by their changing situation and that coordination of attention is predictive of implicit and explicit learning. Chapter 7 presents evidence that joint attention does not require gaze following and that attentional coordination is predictive of learning a manual task. Together, these theoretical and empirical studies suggest a new way of thinking about how humans and other animals live and learn socially, one that is consistent with critical theoretical and philosophical developments that are currently neglected in the literature on social learning.
27
Gargesa, Padmashri. "Reward-driven Training of Random Boolean Network Reservoirs for Model-Free Environments". PDXScholar, 2013. https://pdxscholar.library.pdx.edu/open_access_etds/669.
Texto completoResumen
Reservoir Computing (RC) is an emerging machine learning paradigm where a fixed kernel, built from a randomly connected "reservoir" with sufficiently rich dynamics, is capable of expanding the problem space in a non-linear fashion to a higher dimensional feature space. These features can then be interpreted by a linear readout layer that is trained by a gradient descent method. In comparison to traditional neural networks, only the output layer needs to be trained, which leads to a significant computational advantage. In addition, the short term memory of the reservoir dynamics has the ability to transform a complex temporal input state space to a simple non-temporal representation. Adaptive real-time systems are multi-stage decision problems that can be used to train an agent to achieve a preset goal by performing an optimal action at each timestep. In such problems, the agent learns through continuous interactions with its environment. Conventional techniques to solving such problems become computationally expensive or may not converge if the state-space being considered is large, partially observable, or if short term memory is required in optimal decision making. The objective of this thesis is to use reservoir computers to solve such goal-driven tasks, where no error signal can be readily calculated to apply gradient descent methodologies. To address this challenge, we propose a novel reinforcement learning approach in combination with reservoir computers built from simple Boolean components. Such reservoirs are of interest because they have the potential to be fabricated by self-assembly techniques. We evaluate the performance of our approach in both Markovian and non-Markovian environments. We compare the performance of an agent trained through traditional Q-Learning. We find that the reservoir-based agent performs successfully in these problem contexts and even performs marginally better than Q-Learning agents in certain cases. Our proposed approach allows to retain the advantage of traditional parameterized dynamic systems in successfully modeling embedded state-space representations while eliminating the complexity involved in training traditional neural networks. To the best of our knowledge, our method of training a reservoir readout layer through an on-policy boot-strapping approach is unique in the field of random Boolean network reservoirs.
28
Nguyen, Van Duong. "Variational deep learning for time series modelling and analysis : applications to dynamical system identification and maritime traffic anomaly detection". Thesis, Ecole nationale supérieure Mines-Télécom Atlantique Bretagne Pays de la Loire, 2020. http://www.theses.fr/2020IMTA0227.
Texto completoResumen
Ce travail de thèse se focalise sur une classe de méthodes d’apprentissage profond, probabilistes et non-supervisées qui utilisent l’inférence variationnelle pour créer des modèles évolutifs de grande capacité pour ce type de données. Nous présentons deux classes d’apprentissage variationnel profond, puis nous les appliquons à deux problèmes spécifiques liés au domaine maritime. La première application est l’identification de systèmes dynamiques à partir de données bruitées et partiellement observées. Nous introduisons un cadre qui fusionne l’assimilation de données classique et l’apprentissage profond moderne pour retrouver les équations différentielles qui contrôlent la dynamique du système. En utilisant une formulation d’espace d’états, le cadre proposé intègre des composantes stochastiques pour tenir compte des variabilités stochastiques, des erreurs de modèle et des incertitudes de reconstruction. La deuxième application est la surveillance du trafic maritime à l’aide des données AIS. Nous proposons une architecture d’apprentissage profond probabiliste multitâche pouvant atteindre des performances très prometteuses dans différentes tâches liées à la surveillance du trafic maritime, telles que la reconstruction de trajectoire, l’identification du type de navire et la détection d’anomalie, tout en réduisant considérablement la quantité de données à stocker et le temps de calcul. temps. Pour la tâche la plus importante - la détection d’anomalie, nous introduisons un détecteur géospatialisé qui utilise l’apprentissage profond variationnel pour construire une représentation probabiliste des trajectoires AIS, puis détecter les anomalies en jugeant la probabilité de cette trajectoireThis thesis work focuses on a class of unsupervised, probabilistic deep learning methods that use variational inference to create high capacity, scalable models for time series modelling and analysis. We present two classes of variational deep learning, then apply them to two specific problems related to the maritime domain. The first application is the identification of dynamical systems from noisy and partially observed data. We introduce a framework that merges classical data assimilation and modern deep learning to retrieve the differential equations that control the dynamics of the system. Using a state space formulation, the proposed framework embeds stochastic components to account for stochastic variabilities, model errors and reconstruction uncertainties. The second application is maritime traffic surveillance using AIS data. We propose a multitask probabilistic deep learning architecture can achieve state-of-the-art performance in different maritime traffic surveillance related tasks, such as trajectory reconstruction, vessel type identification and anomaly detection, while reducing significantly the amount data to be stored and the calculation time. For the most important task—anomaly detection, we introduce a geospatial detector that uses variational deep learning to builds a probabilistic representation of AIS trajectories, then detect anomalies by judging how likely this trajectory is
29
McKiernan, Erin C. y Diano F. Marrone. "CA1 pyramidal cells have diverse biophysical properties, affected by development, experience, and aging". PEERJ INC, 2017. http://hdl.handle.net/10150/625990.
Texto completoResumen
Neuron types (e.g., pyramidal cells) within one area of the brain are often considered homogeneous, despite variability in their biophysical properties. Here we review literature demonstrating variability in the electrical activity of CA1 hippocampal pyramidal cells (PCs), including responses to somatic current injection, synaptic stimulation, and spontaneous network-related activity. In addition, we describe how responses of CA1 PCs vary with development, experience, and aging, and some of the underlying ionic currents responsible. Finally, we suggest directions that may be the most impactful in expanding this knowledge, including the use of text and data mining to systematically study cellular heterogeneity in more depth; dynamical systems theory to understand and potentially classify neuron firing patterns; and mathematical modeling to study the interaction between cellular properties and network output. Our goals are to provide a synthesis of the literature for experimentalists studying CA1 PCs, to give theorists an idea of the rich diversity of behaviors models may need to reproduce to accurately represent these cells, and to provide suggestions for future research.
30
Rodrigues, Jose H. "The acquisition of pedagogical expertise in dance : a constraints-led approach". Thesis, Queensland University of Technology, 2017. https://eprints.qut.edu.au/104817/1/Jose_Rodrigues_Thesis.pdf.
Texto completoResumen
This thesis explored the acquisition of pedagogical expertise by dance teachers in a tertiary setting. The constraints-led theoretical framework with a qualitative approach was utilised to undertake a retrospective investigation of the potential factors, or constraints, influencing the acquisition of pedagogical expertise in dance. The results identified five themes, which were mapped into the constraints-led theoretical framework as environmental constraints (i.e., mentors, role models and students); task constraints (i.e., rules); and, individual constraints (i.e., needs). The results from this study highlight the potential of understanding constraints within their specific context in order to improve dance teachers’ pedagogical development.
31
Woodbury, Nathan Scott. "Representation and Reconstruction of Linear, Time-Invariant Networks". BYU ScholarsArchive, 2019. https://scholarsarchive.byu.edu/etd/7402.
Texto completoResumen
Network reconstruction is the process of recovering a unique structured representation of some dynamic system using input-output data and some additional knowledge about the structure of the system. Many network reconstruction algorithms have been proposed in recent years, most dealing with the reconstruction of strictly proper networks (i.e., networks that require delays in all dynamics between measured variables). However, no reconstruction technique presently exists capable of recovering both the structure and dynamics of networks where links are proper (delays in dynamics are not required) and not necessarily strictly proper.The ultimate objective of this dissertation is to develop algorithms capable of reconstructing proper networks, and this objective will be addressed in three parts. The first part lays the foundation for the theory of mathematical representations of proper networks, including an exposition on when such networks are well-posed (i.e., physically realizable). The second part studies the notions of abstractions of a network, which are other networks that preserve certain properties of the original network but contain less structural information. As such, abstractions require less a priori information to reconstruct from data than the original network, which allows previously-unsolvable problems to become solvable. The third part addresses our original objective and presents reconstruction algorithms to recover proper networks in both the time domain and in the frequency domain.
32
García, López Gloria Soledad. "La modelización de las experiencias de enseñanza y aprendizaje del curso diseño de la forma en el espacio estructural". Doctoral thesis, Universitat Autònoma de Barcelona, 2017. http://hdl.handle.net/10803/406127.
Texto completo
33
Massé, Pierre-Yves. "Autour De L'Usage des gradients en apprentissage statistique". Thesis, Université Paris-Saclay (ComUE), 2017. http://www.theses.fr/2017SACLS568/document.
Texto completoResumen
Nous établissons un théorème de convergence locale de l'algorithme classique d'optimisation de système dynamique RTRL, appliqué à un système non linéaire. L'algorithme RTRL est un algorithme en ligne, mais il doit maintenir une grande quantités d'informations, ce qui le rend impropre à entraîner des systèmes d'apprentissage de taille moyenne. L'algorithme NBT y remédie en maintenant une approximation aléatoire non biaisée de faible taille de ces informations. Nous prouvons également la convergence avec probabilité arbitrairement proche de un, de celui-ci vers l'optimum local atteint par l'algorithme RTRL. Nous formalisons également l'algorithme LLR et en effectuons une étude expérimentale, sur des données synthétiques. Cet algorithme met à jour de manière adaptive le pas d'une descente de gradient, par descente de gradient sur celui-ci. Il apporte ainsi une réponse partielle au problème de la fixation numérique du pas de descente, dont le choix influence fortement la procédure de descente et qui doit sinon faire l'objet d'une recherche empirique potentiellement longue par le praticienWe prove a local convergence theorem for the classical dynamical system optimization algorithm called RTRL, in a nonlinear setting. The rtrl works on line, but maintains a huge amount of information, which makes it unfit to train even moderately big learning models. The NBT algorithm turns it by replacing these informations by a non-biased, low dimension, random approximation. We also prove the convergence with arbitrarily close to one probability, of this algorithm to the local optimum reached by the RTRL algorithm. We also formalize the LLR algorithm and conduct experiments on it, on synthetic data. This algorithm updates in an adaptive fashion the step size of a gradient descent, by conducting a gradient descent on this very step size. It therefore partially solves the issue of the numerical choice of a step size in a gradient descent. This choice influences strongly the descent and must otherwise be hand-picked by the user, following a potentially long research
34
Molter, Colin. "Storing information through complex dynamics in recurrent neural networks". Doctoral thesis, Universite Libre de Bruxelles, 2005. http://hdl.handle.net/2013/ULB-DIPOT:oai:dipot.ulb.ac.be:2013/211039.
Texto completoResumen
The neural net computer simulations which will be presented here are based on the acceptance of a set of assumptions that for the last twenty years have been expressed in the fields of information processing, neurophysiology and cognitive sciences. First of all, neural networks and their dynamical behaviors in terms of attractors is the natural way adopted by the brain to encode information. Any information item to be stored in the neural net should be coded in some way or another in one of the dynamical attractors of the brain and retrieved by stimulating the net so as to trap its dynamics in the desired item's basin of attraction. The second view shared by neural net researchers is to base the learning of the synaptic matrix on a local Hebbian mechanism. The last assumption is the presence of chaos and the benefit gained by its presence. Chaos, although very simply produced, inherently possesses an infinite amount of cyclic regimes that can be exploited for coding information. Moreover, the network randomly wanders around these unstable regimes in a spontaneous way, thus rapidly proposing alternative responses to external stimuli and being able to easily switch from one of these potential attractors to another in response to any coming stimulus.In this thesis, it is shown experimentally that the more information is to be stored in robust cyclic attractors, the more chaos appears as a regime in the back, erratically itinerating among brief appearances of these attractors. Chaos does not appear to be the cause but the consequence of the learning. However, it appears as an helpful consequence that widens the net's encoding capacity. To learn the information to be stored, an unsupervised Hebbian learning algorithm is introduced. By leaving the semantics of the attractors to be associated with the feeding data unprescribed, promising results have been obtained in term of storing capacity.
Doctorat en sciences appliquées
info:eu-repo/semantics/nonPublished
35
AlZahrani, Saleh Saeed. "Regionally distributed architecture for dynamic e-learning environment (RDADeLE)". Thesis, De Montfort University, 2010. http://hdl.handle.net/2086/3814.
Texto completoResumen
e-Learning is becoming an influential role as an economic method and a flexible mode of study in the institutions of higher education today which has a presence in an increasing number of college and university courses. e-Learning as system of systems is a dynamic and scalable environment. Within this environment, e-learning is still searching for a permanent, comfortable and serviceable position that is to be controlled, managed, flexible, accessible and continually up-to-date with the wider university structure. As most academic and business institutions and training centres around the world have adopted the e-learning concept and technology in order to create, deliver and manage their learning materials through the web, it has become the focus of investigation. However, management, monitoring and collaboration between these institutions and centres are limited. Existing technologies such as grid, web services and agents are promising better results. In this research a new architecture has been developed and adopted to make the e-learning environment more dynamic and scalable by dividing it into regional data grids which are managed and monitored by agents. Multi-agent technology has been applied to integrate each regional data grid with others in order to produce an architecture which is more scalable, reliable, and efficient. The result we refer to as Regionally Distributed Architecture for Dynamic e-Learning Environment (RDADeLE). Our RDADeLE architecture is an agent-based grid environment which is composed of components such as learners, staff, nodes, regional grids, grid services and Learning Objects (LOs). These components are built and organised as a multi-agent system (MAS) using the Java Agent Development (JADE) platform. The main role of the agents in our architecture is to control and monitor grid components in order to build an adaptable, extensible, and flexible grid-based e-learning system. Two techniques have been developed and adopted in the architecture to build LOs' information and grid services. The first technique is the XML-based Registries Technique (XRT). In this technique LOs' information is built using XML registries to be discovered by the learners. The registries are written in Dublin Core Metadata Initiative (DCMI) format. The second technique is the Registered-based Services Technique (RST). In this technique the services are grid services which are built using agents. The services are registered with the Directory Facilitator (DF) of a JADE platform in order to be discovered by all other components. All components of the RDADeLE system, including grid service, are built as a multi-agent system (MAS). Each regional grid in the first technique has only its own registry, whereas in the second technique the grid services of all regional grids have to be registered with the DF. We have evaluated the RDADeLE system guided by both techniques by building a simulation of the prototype. The prototype has a main interface which consists of the name of the system (RDADeLE) and a specification table which includes Number of Regional Grids, Number of Nodes, Maximum Number of Learners connected to each node, and Number of Grid Services to be filled by the administrator of the RDADeLE system in order to create the prototype. Using the RST technique shows that the RDADeLE system can be built with more regional grids with less memory consumption. Moreover, using the RST technique shows that more grid services can be registered in the RDADeLE system with a lower average search time and the search performance is increased compared with the XRT technique. Finally, using one or both techniques, the XRT or the RST, in the prototype does not affect the reliability of the RDADeLE system.
36
Meghnoudj, Houssem. "Génération de caractéristiques à partir de séries temporelles physiologiques basée sur le contrôle optimal parcimonieux : application au diagnostic de maladies et de troubles humains". Electronic Thesis or Diss., Université Grenoble Alpes, 2024. http://www.theses.fr/2024GRALT003.
Texto completoResumen
Dans cette thèse, une nouvelle méthodologie a été proposée pour la génération de caractéristiques à partir de signaux physiologiques afin de contribuer au diagnostic d'une variété de maladies cérébrales et cardiaques. Basée sur le contrôle optimal parcimonieux, la génération de caractéristiques dynamiques parcimonieuses (SDF) s'inspire du fonctionnement du cerveau. Le concept fondamental de la méthode consiste à décomposer le signal de manière parcimonieuse en modes dynamiques qui peuvent être activés et/ou désactivés au moment approprié avec l'amplitude adéquate. Cette décomposition permet de changer le point de vue sur les données en donnant accès à des caractéristiques plus informatives qui sont plus fidèles au concept de production des signaux cérébraux. Néanmoins, la méthode reste générique et polyvalente puisqu'elle peut être appliquée à un large éventail de signaux. Les performances de la méthode ont été évaluées sur trois problématiques en utilisant des données réelles accessibles publiquement, en abordant des scénarios de diagnostic liés à : (1) la maladie de Parkinson, (2) la schizophrénie et (3) diverses maladies cardiaques. Pour les trois applications, les résultats sont très concluants, puisqu'ils sont comparables aux méthodes de l'état de l'art tout en n'utilisant qu'un petit nombre de caractéristiques (une ou deux pour les applications sur le cerveau) et un simple classifieur linéaire suggérant la robustesse et le bien-fondé des résultats. Il convient de souligner qu'une attention particulière a été accordée à l'obtention de résultats cohérents et significatifs avec une explicabilité sous-jacenteIn this thesis, a novel methodology for features generation from physiological signals (EEG, ECG) has been proposed that is used for the diagnosis of a variety of brain and heart diseases. Based on sparse optimal control, the generation of Sparse Dynamical Features (SDFs) is inspired by the functioning of the brain. The method's fundamental concept revolves around sparsely decomposing the signal into dynamical modes that can be switched on and off at the appropriate time instants with the appropriate amplitudes. This decomposition provides a new point of view on the data which gives access to informative features that are faithful to the brain functioning. Nevertheless, the method remains generic and versatile as it can be applied to a wide range of signals. The methodology's performance was evaluated on three use cases using openly accessible real-world data: (1) Parkinson's Disease, (2) Schizophrenia, and (3) various cardiac diseases. For all three applications, the results are highly conclusive, achieving results that are comparable to the state-of-the-art methods while using only few features (one or two for brain applications) and a simple linear classifier supporting the significance and reliability of the findings. It's worth highlighting that special attention has been given to achieving significant and meaningful results with an underlying explainability
37
Alkhuraiji, Samar. "Dynamic adaptive e-learning system". Thesis, University of Manchester, 2016. https://www.research.manchester.ac.uk/portal/en/theses/dynamic-adaptive-elearning-system(874f7e52-37ab-4454-886f-e98a53ade162).html.
Texto completoResumen
Learning management systems are widely used in educational organizations and universities to deliver self-paced online courses. Furthermore, educational theories have suggested that providing learners with learning material suitable for their learning styles may affect their learning performance. Learners with different individual traits, levels of knowledge, backgrounds, and characteristics are using these learning systems to enhance their learning understanding. This study is concerned with personalizing learning environments based on each learner’s individual needs by designing and developing intelligent adaptive e-learning management systems. These systems behave according to the data collected in a ‘learner model’ from the learner to provide accurate learning material that adapts to learners’ needs by changing the learning environment rapidly based on the learners’ learning requirements and their learning styles. A dynamic adaptive e-learning system (DAELS) is proposed. The idea is to build an algorithm that can quickly understand an individual learner’s learning styles. We propose the Similarity algorithm, which aims to adapt to the student’s learning styles by taking advantage of the experience of previous students that used the same system and studied the same course. This algorithm presents the content to each student according to predictions of his/her preferred learning styles. These predictions can change during a student’s progress and response to the presentation. The ID3 machine learning method was used and integrated into our Similarity algorithm. Such a method can search learners' databases efficiently and quickly by classifying learners based on their attributes. Methods and associated techniques that address these issues by use of Felder and Silverman Learning Styles Model (FSLSM) have been developed and can be built into Moodle, the learning management system, as an integral component. We then conducted experiments on students to evaluate the flexibility of the DAELS and its effect on students’ learning performance. An experiment was designed and implemented to validate the proposed approach’s reliability and performance on learners’ scores. The proposed DAELS was compared with a static adaptive e-learning system (SAELS) and a non-adaptive e-learning system (non-AELS). The results of the empirical experiment demonstrate the effectiveness of using DAELS on student performance. On average, the dynamic adaptive group had an average increase of 60% in the post-test from pre-test, whereas the average score of the static group increased 32%, and the control group had an average increase of 8%. The results reveal that the dynamic group had the highest average scores in the post-test, and the control group had the lowest average increase in scores. The findings indicate that the developed Similarity algorithm, implemented in our DAELS for personalising learning content presentation according to students’ learning styles, is appropriate in e-learning systems and can enhance learning quality.
38
Barfuss, Wolfram. "Learning dynamics and decision paradigms in social-ecological dilemmas". Doctoral thesis, Humboldt-Universität zu Berlin, 2019. http://dx.doi.org/10.18452/20127.
Texto completoResumen
Kollektives Handeln ist erforderlich um nachhaltige Entwicklungspfade in gekoppelten sozial-ökologischen Systemen zu erschließen, fernab von gefährlichen Kippelementen. Ohne anderen Modellierungsprinzipien ihren Nutzen abzuerkennen, schlägt diese Dissertation die Agent-Umwelt Schnittstelle als die mathematische Grundlage für das Modellieren sozial-ökologischer Systeme vor.
Zuerst erweitert diese Arbeit eine Methode aus der Literatur der statistischen Physik über Lerndynamiken, um einen deterministischen Grenzübergang von etablierten Verstärkungslernalgorithmen aus der Forschung zu künstlicher Intelligenz herzuleiten. Die resultierenden Lerndynamiken zeigen eine große Bandbreite verschiedener dynamischer Regime wie z.B. Fixpunkte, Grenzzyklen oder deterministisches Chaos.
Zweitens werden die hergeleiteten Lerngleichungen auf eine neu eingeführte Umwelt, das Ökologisches Öffentliches Gut, angewendet,. Sie modelliert ein gekoppeltes sozial-ökologisches Dilemma und erweitert damit etablierte soziale Dilemmaspiele um ein ökologisches Kippelement. Bekannte theoretische und empirische Ergebnisse werden reproduziert und neuartige, qualitativ verschiedene Parameterregime aufgezeigt, darunter eines, in dem diese belohnungsoptimierenden Lern-Agenten es vorziehen, gemeinsam unter einem Kollaps der Umwelt zu leiden, als in einer florierenden Umwelt zu kooperieren.
Drittens stellt diese Arbeit das Optimierungsparadigma der Lern-Agenten in Frage. Die drei Entscheidungsparadimen ökonomischen Optimierung, Nachhaltigkeit und Sicherheit werden systematisch miteinander verglichen, während sie auf das Management eines umweltlichen Kippelements angewendet werden. Es wird gezeigt, dass kein Paradigma garantiert, Anforderungen anderer Paradigmen zu erfüllen, sowie dass das Fehlen eines Meisterparadigmas von besonderer Bedeutung für das Klimasystem ist, da dieses sich am Rand zwischen Parameterbereichen befinden kann, wo ökonomische Optimierung weder nachhaltig noch sicher wird.Collective action is required to enter sustainable development pathways in coupled social-ecological systems, safely away from dangerous tipping elements. Without denying the usefulness of other model design principles, this thesis proposes the agent-environment interface as the mathematical foundation for the design of social-ecological system models. First, this work refines techniques from the statistical physics literature on learning dynamics to derive a deterministic limit of established reinforcement learning algorithms from artificial intelligence research. Illustrations of the resulting learning dynamics reveal a wide range of different dynamical regimes, such as fixed points, periodic orbits and deterministic chaos. Second, the derived multi-state learning equations are applied to a newly introduced environment, the Ecological Public Good. It models a coupled social-ecological dilemma, extending established repeated social dilemma games by an ecological tipping element. Known theoretical and empirical results are reproduced and novel qualitatively different parameter regimes are discovered, including one in which these reward-optimizing agents prefer to collectively suffer in environmental collapse rather than cooperating in a prosperous environment. Third, this thesis challenges the reward optimizing paradigm of the learning equations. It presents a novel formal comparison of the three decision paradigms of economic optimization, sustainability and safety for the governance of an environmental tipping element. It is shown that no paradigm guarantees fulfilling requirements imposed by another paradigm. Further, the absence of a master paradigm is shown to be of special relevance for governing the climate system, since the latter may reside at the edge between parameter regimes where economic welfare optimization becomes neither sustainable nor safe.
39
Yen, Jerome Chih-Hung. "Stability and learning in dynamic market systems". Diss., The University of Arizona, 1992. http://hdl.handle.net/10150/185843.
Texto completoResumen
An Oligopoly is a special type of market system in which the number of producers is small and the interactions among them are significant. Since the interactions are significant, in order to reach higher profit it is very important for a producer to set up a prediction scheme to predict the decisions of the competitors and a good long-term strategy to occupy greater market share. To model and solve such problems, theoretical studies, field studies, and laboratory experiments (which include computer simulations), are the three major approaches. In this study, theoretical approach was used to develop four prediction schemes and study their stability conditions. Then laboratory experiments were conducted to study the decisions of the human subjects to identify the uses of these prediction schemes. The results of the experiments provided many important messages. In these experiments, not only the prediction schemes that developed by the theoretical approach have been actually used, but also from the strategies that developed by the experiment participants I saw the competitions have moved from the technical level to the psychological level. Based on the findings, I proposed some guidelines to develop a good decision model.
40
North, Ben. "Learning dynamical models for visual tracking". Thesis, University of Oxford, 1998. http://ora.ox.ac.uk/objects/uuid:6ed12552-4c30-4d80-88ef-7245be2d8fb8.
Texto completoResumen
Using some form of dynamical model in a visual tracking system is a well-known method for increasing robustness and indeed performance in general. Often, quite simple models are used and can be effective, but prior knowledge of the likely motion of the tracking target can often be exploited by using a specially-tailored model. Specifying such a model by hand, while possible, is a time-consuming and error-prone process. Much more desirable is for an automated system to learn a model from training data. A dynamical model learnt in this manner can also be a source of useful information in its own right, and a set of dynamical models can provide discriminatory power for use in classification problems. Methods exist to perform such learning, but are limited in that they assume the availability of 'ground truth' data. In a visual tracking system, this is rarely the case. A learning system must work from visual data alone, and this thesis develops methods for learning dynamical models while explicitly taking account of the nature of the training data --- they are noisy measurements. The algorithms are developed within two tracking frameworks. The Kalman filter is a simple and fast approach, applicable where the visual clutter is limited. The recently-developed Condensation algorithm is capable of tracking in more demanding situations, and can also employ a wider range of dynamical models than the Kalman filter, for instance multi-mode models. The success of the learning algorithms is demonstrated experimentally. When using a Kalman filter, the dynamical models learnt using the algorithms presented here produce better tracking when compared with those learnt using current methods. Learning directly from training data gathered using Condensation is an entirely new technique, and experiments show that many aspects of a multi-mode system can be successfully identified using very little prior information. Significant computational effort is required by the implementation of the methods, and there is scope for improvement in this regard. Other possibilities for future work include investigation of the strong links this work has with learning problems in other areas. Most notable is the study of the 'graphical models' commonly used in expert systems, where the ideas presented here promise to give insight and perhaps lead to new techniques.
41
He, Haibo. "Dynamically Self-reconfigurable Systems for Machine Intelligence". Ohio University / OhioLINK, 2006. http://rave.ohiolink.edu/etdc/view?acc_num=ohiou1152717376.
Texto completo
42
Bondorowicz, Stefan. "Adaptive control of complex dynamic systems". Thesis, University of Oxford, 1991. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.302787.
Texto completo
43
Dalla, Libera Alberto. "Learning algorithms for robotics systems". Doctoral thesis, Università degli studi di Padova, 2019. http://hdl.handle.net/11577/3422839.
Texto completoResumen
Robotics systems are now increasingly widespread in our day-life. For instance, robots have been successfully used in several fields, like, agriculture, construction, defense, aerospace, and hospitality. However, there are still several issues to be addressed for allowing the large scale deployment of robots. Issues related to security, and manufacturing and operating costs are particularly relevant. Indeed, differently from industrial applications, service robots should be cheap and capable of operating in unknown, or partially-unknown environments, possibly with minimal human intervention. To deal with these challenges, in the last years the research community focused on deriving learning algorithms capable of providing flexibility and adaptability to the robots. In this context, the application of Machine Learning and Reinforcement Learning techniques turns out to be especially useful. In this manuscript, we propose different learning algorithms for robotics systems. In Chapter 2, we propose a solution for learning the geometrical model of a robot directly from data, combining proprioceptive measures with data collected with a 2D camera. Besides testing the accuracy of the kinematic models derived with real experiments, we validate the possibility of deriving a kinematic controller based on the model identified. Instead, in Chapter 3, we address the robot inverse dynamics problem. Our strategy relies on the fact that the robot inverse dynamics is a polynomial function in a particular input space. Besides characterizing the input space, we propose a data-driven solution based on Gaussian Process Regression (GPR). Given the type of each joint, we define a kernel named Geometrically Inspired Polynomial (GIP) kernel, which is given by the product of several polynomial kernels. To cope with the dimensionality of the resulting polynomial, we use a variation of the standard polynomial kernel, named Multiplicative Polynomial kernel, further discussed in Chapter 6. Tests performed on simulated and real environments show that, compared to other data-driven solutions, the GIP kernel-based estimator is more accurate and data-efficient.
In Chapter 4, we propose a proprioceptive collision detection algorithm based on GPR. Compared to other proprioceptive approaches, we closely inspect the robot behaviors in quasi-static configurations, namely, configurations in which joint velocities are null or close to zero. Such configurations are particularly relevant in the Collaborative Robotics context, where humans and robots work side-by-side sharing the same environment. Experimental results performed with a UR10 robot confirm the relevance of the problem and the effectiveness of the proposed solution.
Finally, in Chapter 5, we present MC-PILCO, a model-based policy search algorithm inspired by the PILCO algorithm. As the original PILCO algorithm, MC-PILCO models the system evolution relying on GPR, and improves the control policy minimizing the expected value of a cost function. However, instead of approximating the expected cost by moment matching, MC-PILCO approximates the expected cost with a Monte Carlo particle-based approach; no assumption about the type of GPR model is necessary. Thus, MC-PILCO allows more freedom in designing the GPR models, possibly leading to better models of the system dynamics. Results obtained in a simulated environment show consistent improvements with respect to the original algorithm, both in terms of speed and success rate.
44
CERBONI, BAIARDI LORENZO. "Adaptive models of learning in complex physical and social systems". Doctoral thesis, Urbino, 2016. http://hdl.handle.net/11576/2630552.
Texto completo
45
Tong, Xiao Thomas. "Statistical Learning of Some Complex Systems: From Dynamic Systems to Market Microstructure". Thesis, Harvard University, 2013. http://dissertations.umi.com/gsas.harvard:10917.
Texto completoResumen
A complex system is one with many parts, whose behaviors are strongly dependent on each other. There are two interesting questions about complex systems. One is to understand how to recover the true structure of a complex system from noisy data. The other is to understand how the system interacts with its environment. In this thesis, we address these two questions by studying two distinct complex systems: dynamic systems and market microstructure. To address the first question, we focus on some nonlinear dynamic systems. We develop a novel Bayesian statistical method, Gaussian Emulator, to estimate the parameters of dynamic systems from noisy data, when the data are either fully or partially observed. Our method shows that estimation accuracy is substantially improved and computation is faster, compared to the numerical solvers. To address the second question, we focus on the market microstructure of hidden liquidity. We propose some statistical models to explain the hidden liquidity under different market conditions. Our statistical results suggest that hidden liquidity can be reliably predicted given the visible state of the market.Statistics
46
McGarity, Michael Computer Science & Engineering Faculty of Engineering UNSW. "Heterogeneous representations for reinforcement learning control of dynamic systems". Awarded by:University of New South Wales. School of Computer Science and Engineering, 2004. http://handle.unsw.edu.au/1959.4/19350.
Texto completoResumen
Intelligent agents are designed to interact with, and learn about, their environment so that they can act purposefully towards a goal. One class of problems encountered in building such agents is learning how to respond to dynamic systems with a continuous state space. The goals of this dissertation are to develop a framework for understanding the behaviour of partitioned dynamic systems with continuous underlying state and to translate this framework into algorithms which adaptively form a partition of the continuous space such that the partitioned system is more easily learned and controlled, and such that the control law may be easily explained in intuitive ways. Currently, algorithms which learn a control policy for partitioned continuous state space systems treat the partitioned system as an approximation to a Markov chain. I give conditions for the partitioned system to be a Markov chain, a semi-Markov process and a new class of system, a weak-semi-Markov process. The weak-semi-Markov model is shown to model partitioned dynamic systems with greater economy than other surveyed models. The behaviour of a partitioned state space system in the area around the region boundaries is also considered. I use the theory of sliding surfaces, and some heuristic arguments to recommend region boundary shape and position. The concept of 'staying on the boundary' then becomes a robust and relatively easy subgoal within the control algorithm. The concept of 'reaching the sliding surface' as a subgoal is used as the basis for an intuitive explanation of the learnt controller. I present an algorithm based on this concept which explains the behaviour of a learnt controller in ways not previously available to a machine learning algorithms. Finally, the Markov Property and the theory of Sliding Mode Control are used as the basis of a class of recursive algorithms. These algorithms adaptively find a partition, and simultaneously use this partition in conjunction with one of five reinforcement learning algorithms to find a control policy based on that partition. This technique is shown to work very well in learning, controlling and explaining a variety of physical systems, from a monorail to a container crane.
47
Yang, Shanhu. "An Adaptive Prognostic Methodology and System Framework for Engineering Systems under Dynamic Working Regimes". University of Cincinnati / OhioLINK, 2016. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1455209450.
Texto completo
48
JÃnior, Amauri Holanda de Souza. "Regional Models and Minimal Learning Machines for Nonlinear Dynamical System Identification". Universidade Federal do CearÃ, 2014. http://www.teses.ufc.br/tde_busca/arquivo.php?codArquivo=14269.
Texto completoResumen
This thesis addresses the problem of identifying nonlinear dynamic systems from a machine learning perspective. In this context, very little is assumed to be known about the system under investigation, and the only source of information comes from input/output measurements on the system. It corresponds to the black-box modeling approach. Numerous strategies and models have been proposed over the last decades in the machine learning field and applied to modeling tasks in a straightforward way. Despite of this variety, the methods can be roughly categorized into global and local modeling approaches. Global modeling consists in fitting a single regression model to the available data, using the whole set of input and output observations. On the other side of the spectrum stands the local modeling approach, in which the input space is segmented into several small partitions and a specialized regression model is fit to each partition.
The first contribution of the thesis is a novel supervised global learning model, the Minimal Learning Machine (MLM). Learning in MLM consists in building a linear mapping between input and output distance matrices and then estimating the nonlinear response from the geometrical configuration of the output points. Given its general formulation, the Minimal Learning Machine is inherently capable of operating on nonlinear regression problems as well as on multidimensional response spaces. Naturally, its characteristics make the MLM able to tackle the system modeling problem.
The second significant contribution of the thesis represents a different modeling paradigm, called Regional Modeling (RM), and it is motivated by the parsimonious principle. Regional models stand between the global and local modeling approaches. The proposal consists of a two-level clustering approach in which we first partition the input space using the Self-Organizing Map (SOM), and then perform clustering over the prototypes of the trained SOM. After that, regression models are built over the clusters of SOM prototypes, or regions in the input space.
Even though the proposals of the thesis can be thought as quite general regression or supervised learning models, the performance assessment is carried out in the context of system identification. Comprehensive performance evaluation of the proposed models on synthetic and real-world datasets is carried out and the results compared to those achieved by standard global and local models. The experiments illustrate that the proposed methods achieve accuracies that are comparable to, and even better than, more traditional machine learning methods thus offering a valid alternative to such approaches.This thesis addresses the problem of identifying nonlinear dynamic systems from a machine learning perspective. In this context, very little is assumed to be known about the system under investigation, and the only source of information comes from input/output measurements on the system. It corresponds to the black-box modeling approach. Numerous strategies and models have been proposed over the last decades in the machine learning field and applied to modeling tasks in a straightforward way. Despite of this variety, the methods can be roughly categorized into global and local modeling approaches. Global modeling consists in fitting a single regression model to the available data, using the whole set of input and output observations. On the other side of the spectrum stands the local modeling approach, in which the input space is segmented into several small partitions and a specialized regression model is fit to each partition. The first contribution of the thesis is a novel supervised global learning model, the Minimal Learning Machine (MLM). Learning in MLM consists in building a linear mapping between input and output distance matrices and then estimating the nonlinear response from the geometrical configuration of the output points. Given its general formulation, the Minimal Learning Machine is inherently capable of operating on nonlinear regression problems as well as on multidimensional response spaces. Naturally, its characteristics make the MLM able to tackle the system modeling problem. The second significant contribution of the thesis represents a different modeling paradigm, called Regional Modeling (RM), and it is motivated by the parsimonious principle. Regional models stand between the global and local modeling approaches. The proposal consists of a two-level clustering approach in which we first partition the input space using the Self-Organizing Map (SOM), and then perform clustering over the prototypes of the trained SOM. After that, regression models are built over the clusters of SOM prototypes, or regions in the input space. Even though the proposals of the thesis can be thought as quite general regression or supervised learning models, the performance assessment is carried out in the context of system identification. Comprehensive performance evaluation of the proposed models on synthetic and real-world datasets is carried out and the results compared to those achieved by standard global and local models. The experiments illustrate that the proposed methods achieve accuracies that are comparable to, and even better than, more traditional machine learning methods thus offering a valid alternative to such approaches.
49
Souza, Júnior Amauri Holanda de. "Regional models and minimal learning machines for nonlinear dynamical system identification". reponame:Repositório Institucional da UFC, 2014. http://www.repositorio.ufc.br/handle/riufc/12481.
Texto completoResumen
SOUZA JUNIOR, A. H. Regional models and minimal learning machines for nonlinear dynamical system identification. 2014. 116 f. Tese (Doutorado em Engenharia de Teleinformática) – Centro de Tecnologia, Universidade Federal do Ceará, Fortaleza, 2014.Submitted by Marlene Sousa (mmarlene@ufc.br) on 2015-05-26T13:38:05Z No. of bitstreams: 1 2014_dis_ahsouzajunior.pdf: 5675945 bytes, checksum: da4cd07b3287237a51c36e519d0cae14 (MD5)
Approved for entry into archive by Marlene Sousa(mmarlene@ufc.br) on 2015-05-27T19:40:24Z (GMT) No. of bitstreams: 1 2014_dis_ahsouzajunior.pdf: 5675945 bytes, checksum: da4cd07b3287237a51c36e519d0cae14 (MD5)
Made available in DSpace on 2015-05-27T19:40:24Z (GMT). No. of bitstreams: 1 2014_dis_ahsouzajunior.pdf: 5675945 bytes, checksum: da4cd07b3287237a51c36e519d0cae14 (MD5) Previous issue date: 2014-10-31
This thesis addresses the problem of identifying nonlinear dynamic systems from a machine learning perspective. In this context, very little is assumed to be known about the system under investigation, and the only source of information comes from input/output measurements on the system. It corresponds to the black-box modeling approach. Numerous strategies and models have been proposed over the last decades in the machine learning field and applied to modeling tasks in a straightforward way. Despite of this variety, the methods can be roughly categorized into global and local modeling approaches. Global modeling consists in fitting a single regression model to the available data, using the whole set of input and output observations. On the other side of the spectrum stands the local modeling approach, in which the input space is segmented into several small partitions and a specialized regression model is fit to each partition. The first contribution of the thesis is a novel supervised global learning model, the Minimal Learning Machine (MLM). Learning in MLM consists in building a linear mapping between input and output distance matrices and then estimating the nonlinear response from the geometrical configuration of the output points. Given its general formulation, the Minimal Learning Machine is inherently capable of operating on nonlinear regression problems as well as on multidimensional response spaces. Naturally, its characteristics make the MLM able to tackle the system modeling problem. The second significant contribution of the thesis represents a different modeling paradigm, called Regional Modeling (RM), and it is motivated by the parsimonious principle. Regional models stand between the global and local modeling approaches. The proposal consists of a two-level clustering approach in which we first partition the input space using the Self-Organizing Map (SOM), and then perform clustering over the prototypes of the trained SOM. After that, regression models are built over the clusters of SOM prototypes, or regions in the input space. Even though the proposals of the thesis can be thought as quite general regression or supervised learning models, the performance assessment is carried out in the context of system identification. Comprehensive performance evaluation of the proposed models on synthetic and real-world datasets is carried out and the results compared to those achieved by standard global and local models. The experiments illustrate that the proposed methods achieve accuracies that are comparable to, and even better than, more traditional machine learning methods thus offering a valid alternative to such approaches
50
Kapp, Marcelo Nepomoceno. "Dynamic optimization of classification systems for adaptive incremental learning". Mémoire, École de technologie supérieure, 2010. http://espace.etsmtl.ca/270/1/KAPP_Marcelo_Nepomoceno.pdf.
Texto completoResumen
Lors de I'arrivée de nouvelles données, un systeme d'apprentissage incremental se met a jour
automatiquement sans reexaminer les anciennes donnees. Lors d'un apprentissage incremental,
les parametres des systemes de classification ne sont plus consideres comme invariants
puisqu'ils peuvent evolucr en fonction des donnees entrantes. Ces changemcnts causent dcs
variations dans I'ajustement des parametres du systeme de classification. Si ces variations sont
negligees, la performance finale d'un tel systeme pent etre ulterieurement compromise. De tcls
systemes, adaptes au probleme de classification, sont tres utiles a des fins industrielles ou militaires
car ceux-ci sont a la fois rapides d'execution et peu gourmands en memoire. On observe
en consequence un interet grandissant a I'elaboration de tels systemes.
L'objectif principal de cette these est de developper un systeme capable de s'adapter de fa^on
incrementale a I'arrivee de nouvelles donnees, de suivrc et d'analyscr dynamiqucment les
parametres du systeme optimal pour ainsi pcrmcttrc son adaptation automatique a de nouvelles
situations. Pour ce faire, nous commen9ons par aborder le probleme de la selection optimale
des classificateurs en fonction du temps. Nous proposons une architecture qui combine la
puissance de la theorie de I'intelligence des essaims avec la methode plus conventionnelle de
recherche par grilles.
Des solutions potentielles sont progressivement identifices et mises en evidence pour des bases
de donnees graduellement mises a jour. L'idee principale ici est de considerer I'ajustement
des parametres du classificateur comme un probleme d'optimisation dynamique dependant des
donnees presentees au systeme de maniere continue. En particulier, nous avons montre que si
I'on cherchait a elaborer un classificateur SVM (Support Vector Machines) efficace a partir de
sources de donnees differentes, graduelles ou en series, mieux valait considerer le processus de
selection de modeles comme un processus dynamique qui pent evoluer et changer Ainsi, les
differentes solutions sont adaptees au fil du temps en fonction revolution des connaissances
accessibles sur le probleme de classifications et de I'incertitude sur les donnees.
Ensuite, nous etudions aussi des mesures pour revaluation et la selection d'ensembles de classificateurs
composes de SVMs. Les mesures employees sont basees sur les theories de la
diversite et la marge communement utilisees pour expliquer la performance des ensembles de
classificateurs. Cette etude revele des informations precieuses pour I'elaboration de mesures
de confiance pouvant servir pour la selection des ensembles de classificateurs.
Finalement, la contribution majeure de cette these est une approche d'optimisation dynamique
qui realise un apprentissage incremental et adaptatif en suivant, faisant evoluer et corabinant
V
les hypotheses d'optima en fonction du temps. L'approche fait usage de concepts issus de differentes
theories experimentales, telles quti I'optiraisation dynamique de particules d'essaims,
les classificateurs SVM incrementaux, la detection de changement et la selection dynamique
d'ensembles a partir de niveaux de confiance des classificateurs. Des experiences menees sur
des bases de donnees synthetiques et reelles montrent que I'approche proposee surpasse les
autres methodes de classification souvent utilisees dans des scenarios d'apprentissage incremental.