Artykuły w czasopismach na temat „Learning with pre-built models”
Kliknij ten link, aby zobaczyć inne rodzaje publikacji na ten temat: Learning with pre-built models.
Utwórz poprawne odniesienie w stylach APA, MLA, Chicago, Harvard i wielu innych
Sprawdź 50 najlepszych artykułów w czasopismach naukowych na temat „Learning with pre-built models”.
Przycisk „Dodaj do bibliografii” jest dostępny obok każdej pracy w bibliografii. Użyj go – a my automatycznie utworzymy odniesienie bibliograficzne do wybranej pracy w stylu cytowania, którego potrzebujesz: APA, MLA, Harvard, Chicago, Vancouver itp.
Możesz również pobrać pełny tekst publikacji naukowej w formacie „.pdf” i przeczytać adnotację do pracy online, jeśli odpowiednie parametry są dostępne w metadanych.
Przeglądaj artykuły w czasopismach z różnych dziedzin i twórz odpowiednie bibliografie.
1
Fauszt, Tibor, László Bognár i Ágnes Sándor. "Increasing the Prediction Power of Moodle Machine Learning Models with Self-defined Indicators". International Journal of Emerging Technologies in Learning (iJET) 16, nr 24 (21.12.2021): 23–39. http://dx.doi.org/10.3991/ijet.v16i24.23923.
Pełny tekst źródłaStreszczenie:
Starting with version 3.4 of Moodle, it has been possible to build educational ML models using predefined indicators in the Analytics API. These models can be used primarily to identify students at risk of failure. Our research shows that the goodness and predictability of models built using predefined core indicators in the API lags far behind the generally acceptable level. Moodle is an open-source system, which on the one hand allows the analysis of algorithms, and on the oth-er hand its modification and further development. Utilizing the openness of the system, we examined the calculation algorithm of the core indicators, and then, based on the experience, we built new models with our own indicators. Our re-sults show that the goodness of models built on a given course can be significant-ly improved. In the article, we discuss the development process in detail and pre-sent the results achieved.
2
Erian, Karim H., Pedro H. Regalado i James M. Conrad. "Missing data handling for machine learning models". IAES International Journal of Robotics and Automation (IJRA) 10, nr 2 (1.06.2021): 123. http://dx.doi.org/10.11591/ijra.v10i2.pp123-132.
Pełny tekst źródłaStreszczenie:
This paper discusses a novel algorithm for solving a missing data problem in the machine learning pre-processing stage. A model built to help lenders evaluate home loans based on numerous factors by learning from available user data, is adopted in this paper as an example. If one of the factors is missing for a person in the dataset, the currently used methods delete the whole entry therefore reducing the size of the dataset and affecting the machine learning model accuracy. The novel algorithm aims to avoid losing entries for missing factors by breaking the dataset into multiple subsets, building a different machine learning model for each subset, then combining the models into one machine learning model. In this manner, the model makes use of all available data and only neglects the missing values. Overall, the new algorithm improved the prediction accuracy by 5% from 93% accuracy to 98% in the home loan example.
3
Cinar, Eyup. "A Sensor Fusion Method Using Transfer Learning Models for Equipment Condition Monitoring". Sensors 22, nr 18 (8.09.2022): 6791. http://dx.doi.org/10.3390/s22186791.
Pełny tekst źródłaStreszczenie:
Sensor fusion is becoming increasingly popular in condition monitoring. Many studies rely on a fusion-level strategy to enable the most effective decision-making and improve classification accuracy. Most studies rely on feature-level fusion with a custom-built deep learning architecture. However, this may limit the ability to use the widely available pre-trained deep learning architectures available to users today. This study proposes a new method for sensor fusion based on concepts inspired by image fusion. The method enables the fusion of multiple and heterogeneous sensors in the time-frequency domain by fusing spectrogram images. The method’s effectiveness is tested with transfer learning (TL) techniques on four different pre-trained convolutional neural network (CNN) based model architectures using an original test environment and data acquisition system. The results show that the proposed sensor fusion technique effectively classifies device faults and the pre-trained TL models enrich the model training capabilities.
4
Lage, Isaac, Emily Chen, Jeffrey He, Menaka Narayanan, Been Kim, Samuel J. Gershman i Finale Doshi-Velez. "Human Evaluation of Models Built for Interpretability". Proceedings of the AAAI Conference on Human Computation and Crowdsourcing 7 (28.10.2019): 59–67. http://dx.doi.org/10.1609/hcomp.v7i1.5280.
Pełny tekst źródłaStreszczenie:
Recent years have seen a boom in interest in interpretable machine learning systems built on models that can be understood, at least to some degree, by domain experts. However, exactly what kinds of models are truly human-interpretable remains poorly understood. This work advances our understanding of precisely which factors make models interpretable in the context of decision sets, a specific class of logic-based model. We conduct carefully controlled human-subject experiments in two domains across three tasks based on human-simulatability through which we identify specific types of complexity that affect performance more heavily than others–trends that are consistent across tasks and domains. These results can inform the choice of regularizers during optimization to learn more interpretable models, and their consistency suggests that there may exist common design principles for interpretable machine learning systems.
5
Yimam, Seid Muhie, Abinew Ali Ayele, Gopalakrishnan Venkatesh, Ibrahim Gashaw i Chris Biemann. "Introducing Various Semantic Models for Amharic: Experimentation and Evaluation with Multiple Tasks and Datasets". Future Internet 13, nr 11 (27.10.2021): 275. http://dx.doi.org/10.3390/fi13110275.
Pełny tekst źródłaStreszczenie:
The availability of different pre-trained semantic models has enabled the quick development of machine learning components for downstream applications. However, even if texts are abundant for low-resource languages, there are very few semantic models publicly available. Most of the publicly available pre-trained models are usually built as a multilingual version of semantic models that will not fit well with the need for low-resource languages. We introduce different semantic models for Amharic, a morphologically complex Ethio-Semitic language. After we investigate the publicly available pre-trained semantic models, we fine-tune two pre-trained models and train seven new different models. The models include Word2Vec embeddings, distributional thesaurus (DT), BERT-like contextual embeddings, and DT embeddings obtained via network embedding algorithms. Moreover, we employ these models for different NLP tasks and study their impact. We find that newly-trained models perform better than pre-trained multilingual models. Furthermore, models based on contextual embeddings from FLAIR and RoBERTa perform better than word2Vec models for the NER and POS tagging tasks. DT-based network embeddings are suitable for the sentiment classification task. We publicly release all the semantic models, machine learning components, and several benchmark datasets such as NER, POS tagging, sentiment classification, as well as Amharic versions of WordSim353 and SimLex999.
6
Wang, Chengyu, Mengli Cheng, Xu Hu i Jun Huang. "EasyASR: A Distributed Machine Learning Platform for End-to-end Automatic Speech Recognition". Proceedings of the AAAI Conference on Artificial Intelligence 35, nr 18 (18.05.2021): 16111–13. http://dx.doi.org/10.1609/aaai.v35i18.18028.
Pełny tekst źródłaStreszczenie:
We present EasyASR, a distributed machine learning platform for training and serving large-scale Automatic Speech Recognition (ASR) models, as well as collecting and processing audio data at scale. Our platform is built upon the Machine Learning Platform for AI of Alibaba Cloud. Its main functionality is to support efficient learning and inference for end-to-end ASR models on distributed GPU clusters. It allows users to learn ASR models with either pre-defined or user-customized network architectures via simple user interface. On EasyASR, we have produced state-of-the-art results over several public datasets for Mandarin speech recognition.
7
Singh, Hrithik, Shambhavi Kaushik, Shruti Talyan i Kartikeya Dwivedi. "Skin Cancer Detection Using Deep Learning techniques". International Journal for Research in Applied Science and Engineering Technology 10, nr 5 (31.05.2022): 4296–305. http://dx.doi.org/10.22214/ijraset.2022.43090.
Pełny tekst źródłaStreszczenie:
Abstract: Skin cancer detection is one of the major prob-lems across the world. Early detection of the skin cancer and its diagnosis is very important for the further treatment of it. Artificial Intelligence has progressed a lot in the field of healthcare and diagnosis and hence skin cancer can also be detected using Machine Leaning and AI. In this research, we have used convolutional neural network for image processing and recognition. The models implemented are Vgg-16, mobilenet, inceptionV3. The paper also reviewed different AI based skin cancer detection models. Here we have used transfer learning method to reuse a pre-trained model also a model from the scratch is also built using CNN blocks. A web app is also featured using HTML, Flask and CSS in which we just have to put the diagnosis image and it will predict the result. Hence, these pre-trained models and a new model from scratch are applied to procure the most optimal model to detect skin cancer using images and web app helps on getting the result at the user end. Thus, the methodology used in this paper if implemented will give improved results of early skin cancer detection using deep learning methods. Index Terms: Skin Cancer, VGG-16, deep learning, convolu-tional neural network, transfer learning.
8
Pedditzi, Maria L., i Marcello Nonnis. "Pre-service Teachers' Representations About Children's Learning: A Pilot Study". Open Psychology Journal 13, nr 1 (13.11.2020): 315–20. http://dx.doi.org/10.2174/1874350102013010315.
Pełny tekst źródłaStreszczenie:
Background:Research on teachers' representations of children's learning is currently ongoing. Social representations are common-sense theories built and shared in everyday interactions. Their analysis can detect the possible differences between teachers’ naïve beliefs and scientific learning theories. Objective: The objective of this pilot study is to analyse the beliefs about children’s learning of a group of teachers. The beliefs will be related to the most acknowledged learning theories. Methods: A mixed methods research was employed to analyse 100 pre-service teachers’ representations of the origins of learning and the psychological processes involved. Results: It emerged from the results that the teachers interviewed consider children’s learning mainly as culturally acquired, which reveals the prevailing constructivist conception of learning. Many pre-service primary school teachers, however, tend to see learning as mere ‘transfer of information’; many pre-service kindergarten teachers perceive learning as ‘behaviour modification’. The most considered psychological aspects are ‘knowledge’ and ‘acquisition’, while emotions are barely considered. Conclusion: Linking implicit theories and disciplinary theories could support pre-service teachers in integrating the theory and the practice of learning so as to understand the way their models influence their educational choices.
9
Zhang, Li, Haimeng Fan, Chengxia Peng, Guozheng Rao i Qing Cong. "Sentiment Analysis Methods for HPV Vaccines Related Tweets Based on Transfer Learning". Healthcare 8, nr 3 (28.08.2020): 307. http://dx.doi.org/10.3390/healthcare8030307.
Pełny tekst źródłaStreszczenie:
The widespread use of social media provides a large amount of data for public sentiment analysis. Based on social media data, researchers can study public opinions on human papillomavirus (HPV) vaccines on social media using machine learning-based approaches that will help us understand the reasons behind the low vaccine coverage. However, social media data is usually unannotated, and data annotation is costly. The lack of an abundant annotated dataset limits the application of deep learning methods in effectively training models. To tackle this problem, we propose three transfer learning approaches to analyze the public sentiment on HPV vaccines on Twitter. One was transferring static embeddings and embeddings from language models (ELMo) and then processing by bidirectional gated recurrent unit with attention (BiGRU-Att), called DWE-BiGRU-Att. The others were fine-tuning pre-trained models with limited annotated data, called fine-tuning generative pre-training (GPT) and fine-tuning bidirectional encoder representations from transformers (BERT). The fine-tuned GPT model was built on the pre-trained generative pre-training (GPT) model. The fine-tuned BERT model was constructed with BERT model. The experimental results on the HPV dataset demonstrated the efficacy of the three methods in the sentiment analysis of the HPV vaccination task. The experimental results on the HPV dataset demonstrated the efficacy of the methods in the sentiment analysis of the HPV vaccination task. The fine-tuned BERT model outperforms all other methods. It can help to find strategies to improve vaccine uptake.
10
Achicanoy, Harold, Deisy Chaves i Maria Trujillo. "StyleGANs and Transfer Learning for Generating Synthetic Images in Industrial Applications". Symmetry 13, nr 8 (16.08.2021): 1497. http://dx.doi.org/10.3390/sym13081497.
Pełny tekst źródłaStreszczenie:
Deep learning applications on computer vision involve the use of large-volume and representative data to obtain state-of-the-art results due to the massive number of parameters to optimise in deep models. However, data are limited with asymmetric distributions in industrial applications due to rare cases, legal restrictions, and high image-acquisition costs. Data augmentation based on deep learning generative adversarial networks, such as StyleGAN, has arisen as a way to create training data with symmetric distributions that may improve the generalisation capability of built models. StyleGAN generates highly realistic images in a variety of domains as a data augmentation strategy but requires a large amount of data to build image generators. Thus, transfer learning in conjunction with generative models are used to build models with small datasets. However, there are no reports on the impact of pre-trained generative models, using transfer learning. In this paper, we evaluate a StyleGAN generative model with transfer learning on different application domains—training with paintings, portraits, Pokémon, bedrooms, and cats—to generate target images with different levels of content variability: bean seeds (low variability), faces of subjects between 5 and 19 years old (medium variability), and charcoal (high variability). We used the first version of StyleGAN due to the large number of publicly available pre-trained models. The Fréchet Inception Distance was used for evaluating the quality of synthetic images. We found that StyleGAN with transfer learning produced good quality images, being an alternative for generating realistic synthetic images in the evaluated domains.
11
Mars, Mourad. "From Word Embeddings to Pre-Trained Language Models: A State-of-the-Art Walkthrough". Applied Sciences 12, nr 17 (1.09.2022): 8805. http://dx.doi.org/10.3390/app12178805.
Pełny tekst źródłaStreszczenie:
With the recent advances in deep learning, different approaches to improving pre-trained language models (PLMs) have been proposed. PLMs have advanced state-of-the-art (SOTA) performance on various natural language processing (NLP) tasks such as machine translation, text classification, question answering, text summarization, information retrieval, recommendation systems, named entity recognition, etc. In this paper, we provide a comprehensive review of prior embedding models as well as current breakthroughs in the field of PLMs. Then, we analyse and contrast the various models and provide an analysis of the way they have been built (number of parameters, compression techniques, etc.). Finally, we discuss the major issues and future directions for each of the main points.
12
An, Yaxin, i Sanket A. Deshmukh. "Machine learning approach for accurate backmapping of coarse-grained models to all-atom models". Chemical Communications 56, nr 65 (2020): 9312–15. http://dx.doi.org/10.1039/d0cc02651d.
Pełny tekst źródłaStreszczenie:
Four different machine learning (ML) regression models: artificial neural network, k-nearest neighbors, Gaussian process regression and random forest were built to backmap coarse-grained models to all-atom models.
13
Pallepati, Manvith. "Network Intrusion Detection System Using Machine Learning with Data Preprocessing and Feature Extraction". International Journal for Research in Applied Science and Engineering Technology 10, nr 6 (30.06.2022): 2360–65. http://dx.doi.org/10.22214/ijraset.2022.44326.
Pełny tekst źródłaStreszczenie:
Abstract: Unauthorized access to a computer network can be discovered by scanning the network traffic for evidence of malicious activity, which is what Network Intrusion Detection (NID) does. However, in this study, we will concentrate on the technology, development, and strategic importance that make up the large field of Network Intrusion Detection (NID). Many new strategies have been created in the last few years to help computer security specialists in protecting a single host or an entire network against unauthorized access, theft, and denial-of-service assaults, which are the primary causes of computer crime. Intrusion Detection is critical for both the military and commercial sectors since it is the most significant study area for the future networks' Information Security. In this paper, a model is being proposed, where the data is preprocessed before training with the algorithms. A study done by comparing with other models shows that, the current model built with Random Forest can outperform other existing models built with ANN when the data is preprocessed. After building model after data pre-processing and feature extraction, we are able to achieve 98.71% accuracy on NSL-KDD dataset.
14
Oh, Seungmin, Sangwon Oh, Hyeju Shin i Jinsul Kim. "Pre-processing optimization for learning power consumption prediction models". Journal of Contents Computing 4, nr 1 (30.06.2022): 431–37. http://dx.doi.org/10.9728/jcc.2022.06.4.1.431.
Pełny tekst źródła
15
Iyer-Raniga, Usha, i Pekka Huovila. "Learning through sharing: beyond the traditional North-South learning models for a circular built environment". IOP Conference Series: Earth and Environmental Science 588 (21.11.2020): 022023. http://dx.doi.org/10.1088/1755-1315/588/2/022023.
Pełny tekst źródła
16
Kadam, Vaibhav, Satish Kumar, Arunkumar Bongale, Seema Wazarkar, Pooja Kamat i Shruti Patil. "Enhancing Surface Fault Detection Using Machine Learning for 3D Printed Products". Applied System Innovation 4, nr 2 (14.05.2021): 34. http://dx.doi.org/10.3390/asi4020034.
Pełny tekst źródłaStreszczenie:
In the era of Industry 4.0, the idea of 3D printed products has gained momentum and is also proving to be beneficial in terms of financial and time efforts. These products are physically built layer-by-layer based on the digital Computer Aided Design (CAD) inputs. Nonetheless, 3D printed products are still subjected to defects due to variation in properties and structure, which leads to deterioration in the quality of printed products. Detection of these errors at each layer level of the product is of prime importance. This paper provides the methodology for layer-wise anomaly detection using an ensemble of machine learning algorithms and pre-trained models. The proposed combination is trained offline and implemented online for fault detection. The current work provides an experimental comparative study of different pre-trained models with machine learning algorithms for monitoring and fault detection in Fused Deposition Modelling (FDM). The results showed that the combination of the Alexnet and SVM algorithm has given the maximum accuracy. The proposed fault detection approach has low experimental and computing costs, which can easily be implemented for real-time fault detection.
17
Ekung, Samuel, i Isaac Odesola. "LEARNING MODELS FOR EFFECTIVE PROPAGATION OF SUSTAINABLE CONSTRUCTION PRACTICES IN THE BUILT ENVIRONMENT". Malaysian Journal of Sustainable Environment 4, nr 1 (30.09.2018): 113. http://dx.doi.org/10.24191/myse.v4i1.5610.
Pełny tekst źródłaStreszczenie:
The drive to achieve sustainable built environment has made learning new skills relating to Sustainable Construction Practices (SCP) imperative. This study investigated whether learning method can improve uptake of SCP through knowledge enhancement. Using survey research strategy, data from 206 construction professionals in Nigeria were collected and analysed. The results revealed that Andragogy and Experiential learning models strongly correlated SCP transfer requirements, and are therefore, appropriate models to embed SCP within existing ethos in the built environment. The study espoused the critical roles of experience and hands-on-project as prerequisite reinforcements for effective learning of SCP.
18
Fahey-Gilmour, J., B. Dawson, P. Peeling, J. Heasman i B. Rogalski. "Multifactorial analysis of factors influencing elite Australian football match outcomes: a machine learning approach". International Journal of Computer Science in Sport 18, nr 3 (1.12.2019): 100–124. http://dx.doi.org/10.2478/ijcss-2019-0020.
Pełny tekst źródłaStreszczenie:
Abstract In Australian football (AF), few studies have assessed combinations of pre- game factors and their relation to game outcomes (win/loss) in multivariable analyses. Further, previous research has mostly been confined to association-based linear approaches and post-game prediction, with limited assessment of predictive machine learning (ML) models in a pre-game setting. Therefore, our aim was to use ML techniques to predict game outcomes and produce a hierarchy of important (win/loss) variables. A total of 152 variables (79 absolute and 73 differentials) were used from the 2013–2018 Australian Football League (AFL) seasons. Various ML models were trained (cross-validation) on the 2013–2017 seasons with the–2018 season used as an independent test set. Model performance varied (66.5-73.3% test set accuracy), although the best model (glmnet – 73.3%) rivalled bookmaker predictions in the same period (70.9%). The glmnet model revealed measures of team quality (a player-based rating and a team-based) in their relative form as the most important variables for prediction. Models that contained in-built feature selection or could model non-linear relationships generally performed better. These findings show that AFL game outcomes can be predicted using ML methods and provide a hierarchy of predictors that maximize the chance of winning.
19
Toba, Hapnes, Hendra Bunyamin, Juan Elisha Widyaya, Christian Wibisono i Lucky Surya Haryadi. "Masking preprocessing in transfer learning for damage building detection". IAES International Journal of Artificial Intelligence (IJ-AI) 12, nr 2 (1.06.2023): 552. http://dx.doi.org/10.11591/ijai.v12.i2.pp552-559.
Pełny tekst źródłaStreszczenie:
The sudden climate change occurring in different places in the world has made disasters more unpredictable than before. In addition, responses are often late due to manual processes that have to be performed by experts. Consequently, major advances in computer vision (CV) have prompted researchers to develop smart models to help these experts. We need a strong image representation model, but at the same time, we also need to prepare for a deep learning environment at a low cost. This research attempts to develop transfer learning models using low-cost masking pre-processing in the experimental building damage (xBD) dataset, a large-scale dataset for advancing building damage assessment. The dataset includes eight types of disasters located in fifteen different countries and spans thousands of square kilometers of satellite images. The models are based on U-Net, i.e., AlexNet, visual geometry group (VGG)-16, and ResNet-34. Our experiments show that ResNet-34 is the best with an F1 score of 71.93%, and an intersection over union (IoU) of 66.72%. The models are built on a resolution of 1,024 pixels and use only first-tier images compared to the state-of-the-art baseline. For future orientations, we believe that the approach we propose could be beneficial to improve the efficiency of deep learning training.
20
Qiao, Yanhua, Xiaolei Zhu i Haipeng Gong. "BERT-Kcr: prediction of lysine crotonylation sites by a transfer learning method with pre-trained BERT models". Bioinformatics 38, nr 3 (13.10.2021): 648–54. http://dx.doi.org/10.1093/bioinformatics/btab712.
Pełny tekst źródłaStreszczenie:
Abstract Motivation As one of the most important post-translational modifications (PTMs), protein lysine crotonylation (Kcr) has attracted wide attention, which involves in important physiological activities, such as cell differentiation and metabolism. However, experimental methods are expensive and time-consuming for Kcr identification. Instead, computational methods can predict Kcr sites in silico with high efficiency and low cost. Results In this study, we proposed a novel predictor, BERT-Kcr, for protein Kcr sites prediction, which was developed by using a transfer learning method with pre-trained bidirectional encoder representations from transformers (BERT) models. These models were originally used for natural language processing (NLP) tasks, such as sentence classification. Here, we transferred each amino acid into a word as the input information to the pre-trained BERT model. The features encoded by BERT were extracted and then fed to a BiLSTM network to build our final model. Compared with the models built by other machine learning and deep learning classifiers, BERT-Kcr achieved the best performance with AUROC of 0.983 for 10-fold cross validation. Further evaluation on the independent test set indicates that BERT-Kcr outperforms the state-of-the-art model Deep-Kcr with an improvement of about 5% for AUROC. The results of our experiment indicate that the direct use of sequence information and advanced pre-trained models of NLP could be an effective way for identifying PTM sites of proteins. Availability and implementation The BERT-Kcr model is publicly available on http://zhulab.org.cn/BERT-Kcr_models/. Supplementary information Supplementary data are available at Bioinformatics online.
21
Kubera, Elżbieta, Agnieszka Kubik-Komar, Krystyna Piotrowska-Weryszko i Magdalena Skrzypiec. "Deep Learning Methods for Improving Pollen Monitoring". Sensors 21, nr 10 (19.05.2021): 3526. http://dx.doi.org/10.3390/s21103526.
Pełny tekst źródłaStreszczenie:
The risk of pollen-induced allergies can be determined and predicted based on data derived from pollen monitoring. Hirst-type samplers are sensors that allow airborne pollen grains to be detected and their number to be determined. Airborne pollen grains are deposited on adhesive-coated tape, and slides are then prepared, which require further analysis by specialized personnel. Deep learning can be used to recognize pollen taxa based on microscopic images. This paper presents a method for recognizing a taxon based on microscopic images of pollen grains, allowing the pollen monitoring process to be automated. In this research, a deep CNN (convolutional neural network) model was built from scratch. Publicly available deep neural network models, pre-trained on image data (not including microscopic pictures), were also used. The results show that even a simple deep learning model produces quite good results when the classification of pollen grain taxa is performed directly from the images. The best deep learning model achieved 97.88% accuracy in the difficult task of recognizing three types of pollen grains (birch, alder, and hazel) with similar structures. The derived models can be used to build a system to support pollen monitoring experts in their work.
22
Anjomshoaa, Amin, i Edward Curry. "Transfer Learning in Smart Environments". Machine Learning and Knowledge Extraction 3, nr 2 (29.03.2021): 318–32. http://dx.doi.org/10.3390/make3020016.
Pełny tekst źródłaStreszczenie:
The knowledge embodied in cognitive models of smart environments, such as machine learning models, is commonly associated with time-consuming and costly processes such as large-scale data collection, data labeling, network training, and fine-tuning of models. Sharing and reuse of these elaborated resources between intelligent systems of different environments, which is known as transfer learning, would facilitate the adoption of cognitive services for the users and accelerate the uptake of intelligent systems in smart building and smart city applications. Currently, machine learning processes are commonly built for intra-organization purposes and tailored towards specific use cases with the assumption of integrated model repositories and feature pools. Transferring such services and models beyond organization boundaries is a challenging task that requires human intervention to find the matching models and evaluate them. This paper investigates the potential of communication and transfer learning between smart environments in order to empower a decentralized and peer-to-peer ecosystem for seamless and automatic transfer of services and machine learning models. To this end, we explore different knowledge types in the context of smart built environments and propose a collaboration framework based on knowledge graph principles for describing the machine learning models and their corresponding dependencies.
23
Khalil, Enas A. H., Enas M. F. El Houby i Hoda K. Mohamed. "Machine Learning for Multilabel Emotion Classification in Arabic Tweets". International Journal of Computer Science and Mobile Computing 11, nr 5 (30.05.2022): 222–33. http://dx.doi.org/10.47760/ijcsmc.2022.v11i05.010.
Pełny tekst źródłaStreszczenie:
Multilabel emotion classification is a high priority because it mimics real-life scenarios in which people display a variety of emotions. The text could express a collection of emotions such as happiness, love, and optimism, or sadness, anger, and pessimism. In this framework, the Arabic tweets data provided by SemEval 2018-Task1, E-c subtask have been first preprocessed through different normalization steps, including stemming, stop word removal, special characters, and digits removal. An emotion lexicon has been built to replace the emotions with their meaning related to emotion classes. A word embedding pre-trained model Aravec has been implemented for the feature extraction process because word embedding performed better in this task than other features such as the N-gram model. In the classification process of our framework, different machine learning techniques have been implemented, including Multi-Layer Perceptron (MLP), Support Vector Machine SVM, K Nearest Neighbor (KNN), Ensemble Random Forest (RF), and Ensemble Extra Tree. The best performance was achieved using MLP, whereas SVM proved to perform best over other Traditional machine learning techniques such as KNN, RF, and Extra tree. Extra tree achieved a multilabel Jaccard accuracy of 26.2%, Nearest Neighbor (KNN) of 37.5%, Ensemble Random Forest (RF) of 29.1%, and SVM accuracy of 46.3%. A neural network model Multi-Layer Perceptron (MLP), achieved an accuracy of 48%. The proposed framework has been compared with different previous machine learning models built for this task; the results obtained by the proposed framework outperform other previous models in most cases.
24
Hammad, Mahmoud, Mohammed Al-Smadi, Qanita Bani Baker i Sa’ad A. Al-Zboon. "Using deep learning models for learning semantic text similarity of Arabic questions". International Journal of Electrical and Computer Engineering (IJECE) 11, nr 4 (1.08.2021): 3519. http://dx.doi.org/10.11591/ijece.v11i4.pp3519-3528.
Pełny tekst źródłaStreszczenie:
<span>Question-answering platforms serve millions of users seeking knowledge and solutions for their daily life problems. However, many knowledge seekers are facing the challenge to find the right answer among similar answered questions and writer’s responding to asked questions feel like they need to repeat answers many times for similar questions. This research aims at tackling the problem of learning the semantic text similarity among different asked questions by using deep learning. Three <span>models are implemented to address the aforementioned problem: i) a supervised-machine learning model using XGBoost trained with pre-defined features, ii) an adapted Siamese-based deep learning recurrent architecture trained with pre-defined</span> features, and iii) a Pre-trained deep bidirectional transformer based on BERT model. Proposed models were evaluated using a reference Arabic dataset from the mawdoo3.com company. Evaluation results show that the BERT-based model outperforms the other two models with an F1=92.<span>99%, whereas the Siamese-based model comes in the second place with F1=89.048%, and finally, the XGBoost as a baseline model achieved the lowest</span> result of F1=86.086%.</span>
25
Oh, Seo Hyun, Min Kang i Youngho Lee. "Protected Health Information Recognition by Fine-Tuning a Pre-training Transformer Model". Healthcare Informatics Research 28, nr 1 (31.01.2022): 16–24. http://dx.doi.org/10.4258/hir.2022.28.1.16.
Pełny tekst źródłaStreszczenie:
Objectives: De-identifying protected health information (PHI) in medical documents is important, and a prerequisite to deidentification is the identification of PHI entity names in clinical documents. This study aimed to compare the performance of three pre-training models that have recently attracted significant attention and to determine which model is more suitable for PHI recognition. Methods: We compared the PHI recognition performance of deep learning models using the i2b2 2014 dataset. We used the three pre-training models—namely, bidirectional encoder representations from transformers (BERT), robustly optimized BERT pre-training approach (RoBERTa), and XLNet (model built based on Transformer-XL)—to detect PHI. After the dataset was tokenized, it was processed using an inside-outside-beginning tagging scheme and WordPiecetokenized to place it into these models. Further, the PHI recognition performance was investigated using BERT, RoBERTa, and XLNet. Results: Comparing the PHI recognition performance of the three models, it was confirmed that XLNet had a superior F1-score of 96.29%. In addition, when checking PHI entity performance evaluation, RoBERTa and XLNet showed a 30% improvement in performance compared to BERT. Conclusions: Among the pre-training models used in this study, XLNet exhibited superior performance because word embedding was well constructed using the two-stream self-attention method. In addition, compared to BERT, RoBERTa and XLNet showed superior performance, indicating that they were more effective in grasping the context.
26
Zhao, Mingjun, Haijiang Wu, Di Niu i Xiaoli Wang. "Reinforced Curriculum Learning on Pre-Trained Neural Machine Translation Models". Proceedings of the AAAI Conference on Artificial Intelligence 34, nr 05 (3.04.2020): 9652–59. http://dx.doi.org/10.1609/aaai.v34i05.6513.
Pełny tekst źródłaStreszczenie:
The competitive performance of neural machine translation (NMT) critically relies on large amounts of training data. However, acquiring high-quality translation pairs requires expert knowledge and is costly. Therefore, how to best utilize a given dataset of samples with diverse quality and characteristics becomes an important yet understudied question in NMT. Curriculum learning methods have been introduced to NMT to optimize a model's performance by prescribing the data input order, based on heuristics such as the assessment of noise and difficulty levels. However, existing methods require training from scratch, while in practice most NMT models are pre-trained on big data already. Moreover, as heuristics, they do not generalize well. In this paper, we aim to learn a curriculum for improving a pre-trained NMT model by re-selecting influential data samples from the original training set and formulate this task as a reinforcement learning problem. Specifically, we propose a data selection framework based on Deterministic Actor-Critic, in which a critic network predicts the expected change of model performance due to a certain sample, while an actor network learns to select the best sample out of a random batch of samples presented to it. Experiments on several translation datasets show that our method can further improve the performance of NMT when original batch training reaches its ceiling, without using additional new training data, and significantly outperforms several strong baseline methods.
27
Reza, Ahmed Wasif, Md Mahamudul Hasan, Nazla Nowrin i Mir Moynuddin Ahmed Shibly. "Pre-trained deep learning models in automatic COVID-19 diagnosis". Indonesian Journal of Electrical Engineering and Computer Science 22, nr 3 (1.06.2021): 1540. http://dx.doi.org/10.11591/ijeecs.v22.i3.pp1540-1547.
Pełny tekst źródłaStreszczenie:
Coronavirus Disease (COVID-19) is a devastating pandemic in the history of mankind. It is a highly contagious flu that can spread from human to human without revealing any symptoms. For being so contagious, detecting patients with it and isolating them has become the primary concern for healthcare professionals. This study presented an alternative way to identify COVID-19 patients by doing an automatic examination of chest X-rays of the patients. To develop such an efficient system, six pre-trained deep learning models were used. Those models were: VGG16, InceptionV3, Xception, DenseNet201, InceptionResNetV2, and EfficientNetB4. Those models were developed on two open-source datasets that have chest X-rays of patients diagnosed with COVID-19. Among the models, EfficientNetB4 achieved better performances on both datasets with 96% and 97% of accuracies. The empirical results were also exemplary. This type of automated system can help us fight this dangerous virus outbreak.
28
Rotman, Guy, i Roi Reichart. "Multi-task Active Learning for Pre-trained Transformer-based Models". Transactions of the Association for Computational Linguistics 10 (2022): 1209–28. http://dx.doi.org/10.1162/tacl_a_00515.
Pełny tekst źródłaStreszczenie:
Abstract Multi-task learning, in which several tasks are jointly learned by a single model, allows NLP models to share information from multiple annotations and may facilitate better predictions when the tasks are inter-related. This technique, however, requires annotating the same text with multiple annotation schemes, which may be costly and laborious. Active learning (AL) has been demonstrated to optimize annotation processes by iteratively selecting unlabeled examples whose annotation is most valuable for the NLP model. Yet, multi-task active learning (MT-AL) has not been applied to state-of-the-art pre-trained Transformer-based NLP models. This paper aims to close this gap. We explore various multi-task selection criteria in three realistic multi-task scenarios, reflecting different relations between the participating tasks, and demonstrate the effectiveness of multi-task compared to single-task selection. Our results suggest that MT-AL can be effectively used in order to minimize annotation efforts for multi-task NLP models.1
29
Baranov, Mykola, Yurii Shcherbyna i Oles Khodych. "Exploit computer vision inpainting approach to boost deep learning models". Vìsnik Nacìonalʹnogo unìversitetu "Lʹvìvsʹka polìtehnìka". Serìâ Ìnformacìjnì sistemi ta merežì 12 (15.12.2022): 1–6. http://dx.doi.org/10.23939/sisn2022.12.001.
Pełny tekst źródłaStreszczenie:
In today’s world, the amount of available information grows exponentially every day. Most of this data is visual data. Correspondingly, the demand for the algorithm of image rent is growing. Traditionally, the first approaches to computer vision problems were classical algorithms without the use of machine learning. Such approaches are limited by many factors. First of all, the conditions imposed on the input images are applied – the shooting angle, lighting, position of objects on the scene, etc. Other classical algorithms cannot meet the needs of modern computer vision problems. Neural network approaches and deep learning models have largely replaced classical programming algorithms. The greatest advantage of deep neural networks in computer vision tasks is not only the possibility of automatically building data processing algorithms that cannot be built in any other way, but also the comprehensiveness of such an approach – actual deep neural networks provide all stages of image processing from start to finish. But. This approach is not always optimal. Training models require a large amount of annotated data to avoid the effect of overfitting such models. In many settings, the conditions have a significant degree of variability, but are limited. In such cases, the combination of both approaches of computer vision is fruitful – pre-processing of the image is performed by classical algorithms, and prediction (classification, object search, etc.) is performed by a neural network. This article noted an example of the use of damaged images in the classification of tasks (in the extreme cases, the percentage of damage reached 60 % of the image area). We have shown in practice that the use of classic approaches for restoration of damaged areas of the image (inpainting) made it possible to increase the final accuracy of the model by up to 10 % compared to the base model trained under identical conditions on the original data.
30
Kim, Gyunyeop, i Sangwoo Kang. "Effective Transfer Learning with Label-Based Discriminative Feature Learning". Sensors 22, nr 5 (4.03.2022): 2025. http://dx.doi.org/10.3390/s22052025.
Pełny tekst źródłaStreszczenie:
The performance of natural language processing with a transfer learning methodology has improved by applying pre-training language models to downstream tasks with a large number of general data. However, because the data used in pre-training are irrelevant to the downstream tasks, a problem occurs in that it learns general features rather than those features specific to the downstream tasks. In this paper, a novel learning method is proposed for embedding pre-trained models to learn specific features of such tasks. The proposed method learns the label features of downstream tasks through contrast learning using label embedding and sampled data pairs. To demonstrate the performance of the proposed method, we conducted experiments on sentence classification datasets and evaluated whether the features of the downstream tasks have been learned through a PCA and a clustering of the embeddings.
31
Verbruggen, Gust, Vu Le i Sumit Gulwani. "Semantic programming by example with pre-trained models". Proceedings of the ACM on Programming Languages 5, OOPSLA (20.10.2021): 1–25. http://dx.doi.org/10.1145/3485477.
Pełny tekst źródłaStreszczenie:
The ability to learn programs from few examples is a powerful technology with disruptive applications in many domains, as it allows users to automate repetitive tasks in an intuitive way. Existing frameworks on inductive synthesis only perform syntactic manipulations, where they rely on the syntactic structure of the given examples and not their meaning. Any semantic manipulations, such as transforming dates, have to be manually encoded by the designer of the inductive programming framework. Recent advances in large language models have shown these models to be very adept at performing semantic transformations of its input by simply providing a few examples of the task at hand. When it comes to syntactic transformations, however, these models are limited in their expressive power. In this paper, we propose a novel framework for integrating inductive synthesis with few-shot learning language models to combine the strength of these two popular technologies. In particular, the inductive synthesis is tasked with breaking down the problem in smaller subproblems, among which those that cannot be solved syntactically are passed to the language model. We formalize three semantic operators that can be integrated with inductive synthesizers. To minimize invoking expensive semantic operators during learning, we introduce a novel deferred query execution algorithm that considers the operators to be oracles during learning. We evaluate our approach in the domain of string transformations: the combination methodology can automate tasks that cannot be handled using either technologies by themselves. Finally, we demonstrate the generality of our approach via a case study in the domain of string profiling.
32
Shadiev, Rustam, Xueying Wang, Ting-Ting Wu i Yueh-Min Huang. "Review of Research on Technology-Supported Cross-Cultural Learning". Sustainability 13, nr 3 (29.01.2021): 1402. http://dx.doi.org/10.3390/su13031402.
Pełny tekst źródłaStreszczenie:
Not many review studies have explored the theoretical foundation of cross-cultural learning or the curricula in the research they were reviewing. Furthermore, some review studies only superficially discussed the methodology and findings of the reviewed articles. To address these issues, we reviewed twenty-three studies on technology-supported cross-cultural learning published between 2014 and 2020. We aimed to summarize and analyze previous research in the following areas: (1) theoretical foundation, (2) curricula, (3) technologies, and (4) methodology and findings. Our results showed that the reviewed studies built their research framework based on diverse theoretical foundations; however, the most frequently used models were Byram’s model and the cultural convergence theory. Curricula had the following main focuses: (a) cross-cultural learning, (b) linguistic skills, and (c) pre-service teacher training. The most frequently used technologies were Skype, e-mail, and blogs. We found that most reviewed studies involved the collection of both qualitative and quantitative data. Finally, most of the reviewed studies reported on the role of technologies in facilitating cross-cultural learning, FL/SL learning, and pre-service teacher training. Based on our findings, several implications along with suggestions were prepared. Our findings demonstrated that results from most studies were positive regarding technological support of cross-cultural learning. Therefore, it is suggested that educators and researchers take these results into consideration when designing future studies on cross-cultural learning. Because many scholars did not report some important information, such as what theoretical foundation they built studies on or participants’ demographics, we suggest that such information needs to be included in their research articles as it can be helpful in informing future studies. We also suggest that participants in future studies use variety of technological tools for supporting communication and content creation during cross-cultural learning.
33
Weng, Rongxiang, Heng Yu, Weihua Luo i Min Zhang. "Deep Fusing Pre-trained Models into Neural Machine Translation". Proceedings of the AAAI Conference on Artificial Intelligence 36, nr 10 (28.06.2022): 11468–76. http://dx.doi.org/10.1609/aaai.v36i10.21399.
Pełny tekst źródłaStreszczenie:
Pre-training and fine-tuning have become the de facto paradigm in many natural language processing (NLP) tasks. However, compared to other NLP tasks, neural machine translation (NMT) aims to generate target language sentences through the contextual representation from the source language counterparts. This characteristic means the optimization objective of NMT is far from that of the universal pre-trained models (PTMs), leading to the standard procedure of pre-training and fine-tuning does not work well in NMT. In this paper, we propose a novel framework to deep fuse the pre-trained representation into NMT, fully exploring the potential of PTMs in NMT. Specifically, we directly replace the randomly initialized Transformer encoder with a pre-trained encoder and propose a layer-wise coordination structure to coordinate PTM and NMT decoder learning. Then, we introduce a partitioned multi-task learning method to fine-tune the pre-trained parameter, reducing the gap between PTM and NMT by progressively learning the task-specific representation. Experimental results show that our approach achieves considerable improvements on WMT14 En2De, WMT14 En2Fr, and WMT16 Ro2En translation benchmarks and outperforms previous work in both autoregressive and non-autoregressive NMT models.
34
Jeon, Joohyeong, Byungohk Rhee i Jinsu Gim. "Melt Temperature Estimation by Machine Learning Model Based on Energy Flow in Injection Molding". Polymers 14, nr 24 (18.12.2022): 5548. http://dx.doi.org/10.3390/polym14245548.
Pełny tekst źródłaStreszczenie:
Highly reliable and accurate melt temperature measurements in the barrel are necessary for stable injection molding. Conventional sheath-type thermocouples are insufficiently responsive for measuring melt temperatures during molding. Herein, machine learning models were built to predict the melt temperature after plasticizing. To supply reliably labeled melt temperatures to the models, an optimized temperature sensor was developed. Based on measured high-quality temperature data, three machine learning models were built. The first model accepted process setting parameters as inputs and was built for comparisons with previous models. The second model accepted additional measured process parameters related to material energy flow during plasticizing. Finally, the third model included the specific heat and part weights reflecting the material energy, in addition to the features of the second model. Thus, the third model outperformed the others, and its loss decreased by more than 70%. Meanwhile, the coefficient of determination increased by about 0.5 more than those of the first model. To reduce the dataset size for new materials, a transfer learning model was built using the third model, which showed a high prediction performance and reliability with a smaller dataset. Additionally, the reliability of the input features to the machine learning models were evaluated by shapley additive explanations (SHAP) analysis.
35
Bulla, Premamayudu, Lakshmipathi Anantha i Subbarao Peram. "Deep Neural Networks with Transfer Learning Model for Brain Tumors Classification". Traitement du Signal 37, nr 4 (10.10.2020): 593–601. http://dx.doi.org/10.18280/ts.370407.
Pełny tekst źródłaStreszczenie:
To investigate the effect of deep neural networks with transfer learning on MR images for tumor classification and improve the classification metrics by building image-level, stratified image-level, and patient-level models. Three thousand sixty-four T1-weighted magnetic resonance (MR) imaging from two hundred thirty-three patient cases of three brain tumors types (meningioma, glioma, and pituitary) were collected and it includes coronal, sagittal and axial views. The average number of brain images of each patient in three views is fourteen in the collected dataset. The classification is performed in a model of cross-trained with a pre-trained InceptionV3 model. Three image-level and one patient-level models are built on the MR imaging dataset. The models are evaluated in classification metrics such as accuracy, loss, precision, recall, kappa, and AUC. The proposed models are validated using four approaches: holdout validation, 10-fold cross-validation, stratified 10-fold cross-validation, and group 10-fold cross-validation. The generalization capability and improvement of the network are tested by using cropped and uncropped images of the dataset. The best results for group 10-fold cross-validation (patient-level) are obtained on the used dataset (ACC=99.82). A deep neural network with transfer learning can be used to classify brain tumors from MR images. Our patient-level network model noted the best results in classification to improve accuracy.
36
Patel, Tatsat R., Muhammad Waqas, Seyyed M. M. J. Sarayi, Zeguang Ren, Cesario V. Borlongan, Rimal Dossani, Elad I. Levy i in. "Revascularization Outcome Prediction for A Direct Aspiration-First Pass Technique (ADAPT) from Pre-Treatment Imaging and Machine Learning". Brain Sciences 11, nr 10 (5.10.2021): 1321. http://dx.doi.org/10.3390/brainsci11101321.
Pełny tekst źródłaStreszczenie:
A direct aspiration-first pass technique (ADAPT) has recently gained popularity for the treatment of large vessel ischemic stroke. Here, we sought to create a machine learning-based model that uses pre-treatment imaging metrics to predict successful outcomes for ADAPT in middle cerebral artery (MCA) stroke cases. In 119 MCA strokes treated by ADAPT, we calculated four imaging parameters—clot length, perviousness, distance from the internal carotid artery (ICA) and angle of interaction (AOI) between clot/catheter. We determined treatment success by first pass effect (FPE), and performed univariate analyses. We further built and validated multivariate machine learning models in a random train-test split (75%:25%) of our data. To test model stability, we repeated the machine learning procedure over 100 randomizations, and reported the average performances. Our results show that perviousness (p = 0.002) and AOI (p = 0.031) were significantly higher and clot length (p = 0.007) was significantly lower in ADAPT cases with FPE. A logistic regression model achieved the highest accuracy (74.2%) in the testing cohort, with an AUC = 0.769. The models had similar performance over the 100 train-test randomizations (average testing AUC = 0.768 ± 0.026). This study provides feasibility of multivariate imaging-based predictors for stroke treatment outcome. Such models may help operators select the most adequate thrombectomy approach.
37
B, Naveen Kumar, C. Akshay, Abhishek S P, Ganadakshaka K i K. Jayanth. "An Automated System for Fruit Adulteration and Fruit Grading". International Journal for Research in Applied Science and Engineering Technology 10, nr 6 (30.06.2022): 1223–27. http://dx.doi.org/10.22214/ijraset.2022.44011.
Pełny tekst źródłaStreszczenie:
Abstract: This research covers deep learning approaches, a supervised machine learning model for fruit illness diagnosis, and a convolutional neural network-based fruit grading system. We employed a Visual Geometry Group (VGG) which is a part of the Convolutional Neural Network (CNN) and it produced an accurate result. Fruit disease detection is a tough task for a manual inspection system, so we have designed a system that detects the fruit disease and grades it. In recent times the machines are incorporated with a high-speed computing hardware device that allows the developers to develop the complex system using different types of machine learning models, and algorithms for better results and near accuracy. Using these advanced models of neural networks, a good model is built for classifying the better fruit. The dataset is taken from kaggle.com, and various sorts of fruit photos were utilized to train and test the model, resulting in an accurate result. Key Terms: Supervised machine learning model, Convolutional Neural Network, Visual Geometry Group, Pre-Processing, Feature Extraction, Classification, Gray Scale Conversion, Noise Removal, Thresholding, and Image Sharpening.
38
Khalifa, Noorain, Leela Sarath Kumar Konda i Rajendra Kristam. "Machine learning-based QSAR models to predict sodium ion channel (Nav 1.5) blockers". Future Medicinal Chemistry 12, nr 20 (październik 2020): 1829–43. http://dx.doi.org/10.4155/fmc-2020-0156.
Pełny tekst źródłaStreszczenie:
Aim: Conventional experimental approaches used for the evaluation of the proarrhythmic potential of compounds in the drug discovery process are expensive and time consuming but an integral element in the safety profile required for a new drug to be approved. The voltage-gated sodium ion channel 1.5 (Nav 1.5), a target known for arrhythmic drugs, causes adverse cardiac complications when the channel is blocked. Results: Machine learning classification and regression models were built to predict the possibility of blocking these channels by small molecules. The finalized models tested with balanced accuracies of 0.88, 0.93 and 0.94 at three thresholds (1, 10 and 30 µmol, respectively). The regression model built to predict the pIC50 of compounds had q2 of 0.84 (root-mean-square error = 0.46). Conclusion: The machine learning models that have been built can act as effective filters to screen out the potentially toxic compounds in the early stages of drug discovery.
39
Alelyani, Salem. "Detection and Evaluation of Machine Learning Bias". Applied Sciences 11, nr 14 (7.07.2021): 6271. http://dx.doi.org/10.3390/app11146271.
Pełny tekst źródłaStreszczenie:
Machine learning models are built using training data, which is collected from human experience and is prone to bias. Humans demonstrate a cognitive bias in their thinking and behavior, which is ultimately reflected in the collected data. From Amazon’s hiring system, which was built using ten years of human hiring experience, to a judicial system that was trained using human judging practices, these systems all include some element of bias. The best machine learning models are said to mimic humans’ cognitive ability, and thus such models are also inclined towards bias. However, detecting and evaluating bias is a very important step for better explainable models. In this work, we aim to explain bias in learning models in relation to humans’ cognitive bias and propose a wrapper technique to detect and evaluate bias in machine learning models using an openly accessible dataset from UCI Machine Learning Repository. In the deployed dataset, the potentially biased attributes (PBAs) are gender and race. This study introduces the concept of alternation functions to swap the values of PBAs, and evaluates the impact on prediction using KL divergence. Results demonstrate females and Asians to be associated with low wages, placing some open research questions for the research community to ponder over.
40
Didkovskyi, Oleksandr, Vladislav Ivanov, Alessio Radice, Monica Papini, Laura Longoni i Alessandra Menafoglio. "A Comparison Between Machine Learning and Functional Geostatistics Approaches for Data-Driven Analyses of Sediment Transport in a Pre-Alpine Stream". Mathematical Geosciences 54, nr 3 (16.03.2022): 467–506. http://dx.doi.org/10.1007/s11004-022-09995-9.
Pełny tekst źródłaStreszczenie:
AbstractThe problem of providing data-driven models for sediment transport in a pre-Alpine stream in Italy is addressed. This study is based on a large set of measurements collected from real pebbles, traced along the stream through radio-frequency identification tags after precipitation events. Two classes of data-driven models based on machine learning and functional geostatistics approaches are proposed and evaluated to predict the probability of movement of single pebbles within the stream. The first class built upon gradient-boosting decision trees allows one to estimate the probability of movement of a pebble based on the pebbles’ geometrical features, river flow rate, location, and subdomain types. The second class is built upon functional kriging, a recent geostatistical technique that allows one to predict a functional profile—that is, the movement probability of a pebble, as a function of the pebbles’ geometrical features or the stream’s flow rate—at unsampled locations in the study area. Although grounded in different perspectives, both models aim to account for two main sources of uncertainty, namely, (1) the complexity of a river’s morphological structure and (2) the highly nonlinear dependence between probability of movement, pebble size and shape, and the stream’s flow rate. The performance of the two methods is extensively compared in terms of classification accuracy. The analyses show that despite the different perspectives, the overall performance is adequate and consistent, which suggests that both approaches can provide modeling frameworks for sediment transport. These data-driven approaches are also compared with physics-based ones that are classically used in the hydrological literature. Finally, the use of the developed models in a bottom-up strategy, which starts with the prediction/classification of a single pebble and then integrates the results into a forecast of the grain-size distribution of mobilized sediments, is discussed.
41
Bechelli, Solene, i Jerome Delhommelle. "Machine Learning and Deep Learning Algorithms for Skin Cancer Classification from Dermoscopic Images". Bioengineering 9, nr 3 (27.02.2022): 97. http://dx.doi.org/10.3390/bioengineering9030097.
Pełny tekst źródłaStreszczenie:
We carry out a critical assessment of machine learning and deep learning models for the classification of skin tumors. Machine learning (ML) algorithms tested in this work include logistic regression, linear discriminant analysis, k-nearest neighbors classifier, decision tree classifier and Gaussian naive Bayes, while deep learning (DL) models employed are either based on a custom Convolutional Neural Network model, or leverage transfer learning via the use of pre-trained models (VGG16, Xception and ResNet50). We find that DL models, with accuracies up to 0.88, all outperform ML models. ML models exhibit accuracies below 0.72, which can be increased to up to 0.75 with ensemble learning. To further assess the performance of DL models, we test them on a larger and more imbalanced dataset. Metrics, such as the F-score and accuracy, indicate that, after fine-tuning, pre-trained models perform extremely well for skin tumor classification. This is most notably the case for VGG16, which exhibits an F-score of 0.88 and an accuracy of 0.88 on the smaller database, and metrics of 0.70 and 0.88, respectively, on the larger database.
42
Li, Y., Y. Tong, K. Lu, S. Yu i J. Qian. "P213 Can artificial intelligence help differentiate inflammatory intestinal diseases in China? Models applying random forest and convolutional neuron network". Journal of Crohn's and Colitis 14, Supplement_1 (styczeń 2020): S247—S248. http://dx.doi.org/10.1093/ecco-jcc/jjz203.342.
Pełny tekst źródłaStreszczenie:
Abstract Background Differentiating between ulcerative colitis (UC), Crohn’s disease (CD) and intestinal tuberculosis (ITB) is challenging under endoscopy. We aimed to realise automatic differential diagnosis among these diseases through machine learning algorithms. Methods A total of 6399 consecutive patients (5128 UC, 875 CD and 396 ITB) who had taken colonoscopy examinations in Peking Union Medical College Hospital from January 2008 to November 2018 was enrolled. The input was the description of the endoscopic image in the form of free-text. Word segmentation and key word infiltration were conducted as data pre-processing. Random forest (RF) and convolutional neural network (CNN) were applied to different disease entities. Three two-class classifiers (UC and CD, UC and ITB, CD and ITB) and a three-class classifier (UC, CD and ITB) were built. Sensitivity/specificity and precision/recall were applied to evaluate the performance of two-class classifiers and the three-class classifier, respectively. Results The classifiers built in this research were well-performed and the CNN had a better performance in general. The RF sensitivities/specificities of UC-CD, UC-ITB and CD-ITB were 0.89/0.84, 0.83/0.82 and 0.72/0.77, while the CNN of CD-ITB was 0.90/0.77. The precision/recall of UC-CD-ITB was 0.97/0.97, 0.65/0.53 and 0.68/0.76 by RF, respectively, and 0.99/0.97,0.87/0.83 and 0.52/0.81 by CNN, respectively. Conclusion Classifiers built by RF and CNN had an excellent performance when classifying UC with CD or ITB. For the differentiation of CD and ITB, high specificity and sensitivity were reached as well. Artificial intelligence through machine learning is very promising in helping unexperienced endoscopists differentiate inflammatory intestinal diseases.
43
Idrissi, Idriss, Mostafa Azizi i Omar Moussaoui. "Accelerating the update of a DL-based IDS for IoT using deep transfer learning". Indonesian Journal of Electrical Engineering and Computer Science 23, nr 2 (1.08.2021): 1059. http://dx.doi.org/10.11591/ijeecs.v23.i2.pp1059-1067.
Pełny tekst źródłaStreszczenie:
<p>Deep learning (DL) models are nowadays broadly applied and have shown outstanding performance in a variety of fields, including our focus topic of "IoTcybersecurity". Deep learning-based intrusion detection system (DL-IDS) models are more fixated and depended on the trained dataset. This poses a problem for these DL-IDS, especially with the known mutation and behavior changes of attacks, which can render them undetected. As a result, the DL-IDShas become outdated. In this work, we present a solution for updating DL-ID Semploying a transfer learning technique that allows us to retrain and fine-tune pre-trained models on small datasets with new attack behaviors. In our experiments, we built CNN-based IDS on the Bot-IoT dataset and updated it on small data from a new dataset named TON-IoT. We obtained promising results in multiple metrics regarding the detection rate and the training between the initial training for the original model and the updated one, in the matter of detecting new attacks behaviors and improving the detection rate for some classes by overcoming the lack of their labeled data.</p>
44
Bassier, M., M. Vergauwen i B. Van Genechten. "AUTOMATED CLASSIFICATION OF HERITAGE BUILDINGS FOR AS-BUILT BIM USING MACHINE LEARNING TECHNIQUES". ISPRS Annals of Photogrammetry, Remote Sensing and Spatial Information Sciences IV-2/W2 (16.08.2017): 25–30. http://dx.doi.org/10.5194/isprs-annals-iv-2-w2-25-2017.
Pełny tekst źródłaStreszczenie:
Semantically rich three dimensional models such as Building Information Models (BIMs) are increasingly used in digital heritage. They provide the required information to varying stakeholders during the different stages of the historic buildings life cyle which is crucial in the conservation process. The creation of as-built BIM models is based on point cloud data. However, manually interpreting this data is labour intensive and often leads to misinterpretations. By automatically classifying the point cloud, the information can be proccesed more effeciently. A key aspect in this automated scan-to-BIM process is the classification of building objects.<br><br> In this research we look to automatically recognise elements in existing buildings to create compact semantic information models. Our algorithm efficiently extracts the main structural components such as floors, ceilings, roofs, walls and beams despite the presence of significant clutter and occlusions. More specifically, Support Vector Machines (SVM) are proposed for the classification. The algorithm is evaluated using real data of a variety of existing buildings. The results prove that the used classifier recognizes the objects with both high precision and recall. As a result, entire data sets are reliably labelled at once. The approach enables experts to better document and process heritage assets.
45
Saraubon, Kobkiat, Nuttapong Wiriyanuruknakon i Natdanai Tangthirasunun. "Flashover Prevention System using IoT and Machine Learning for Transmission and Distribution Lines". International Journal of Interactive Mobile Technologies (iJIM) 15, nr 11 (4.06.2021): 34. http://dx.doi.org/10.3991/ijim.v15i11.20753.
Pełny tekst źródłaStreszczenie:
Flashover on transmission and distribution line insulators occurs when the insulator’s resistance drops to a critical level and causes frequent power outages. Thin layers of dust, salt, and airborne particles, gradually deposited on the surface of insulators, as well as humidity, form an electrolyte which causes flashover. In this paper, a flashover prevention system using IoT technology and machine learning is proposed in order to reduce loss and increase power reliability. The system includes an IoT module, a service and clients. The IoT module prototype was installed at a distribution line pole located in Pracha-utit, Bangkok, Thailand and had collected data for thirty-four months. The data were pre-processed and split for the training process and evaluation. In this study, we built and compared four models including linear regression, polynomial regression, Auto-regressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) models. The results revealed that the LSTM model outperformed (<em>R</em><sup>2</sup>=.931, RMSE= 530.74) the others.
46
Abbas, Ather, Laurie Boithias, Yakov Pachepsky, Kyunghyun Kim, Jong Ahn Chun i Kyung Hwa Cho. "AI4Water v1.0: an open-source python package for modeling hydrological time series using data-driven methods". Geoscientific Model Development 15, nr 7 (8.04.2022): 3021–39. http://dx.doi.org/10.5194/gmd-15-3021-2022.
Pełny tekst źródłaStreszczenie:
Abstract. Machine learning has shown great promise for simulating hydrological phenomena. However, the development of machine-learning-based hydrological models requires advanced skills from diverse fields, such as programming and hydrological modeling. Additionally, data pre-processing and post-processing when training and testing machine learning models are a time-intensive process. In this study, we developed a python-based framework that simplifies the process of building and training machine-learning-based hydrological models and automates the process of pre-processing hydrological data and post-processing model results. Pre-processing utilities assist in incorporating domain knowledge of hydrology in the machine learning model, such as the distribution of weather data into hydrologic response units (HRUs) based on different HRU discretization definitions. The post-processing utilities help in interpreting the model's results from a hydrological point of view. This framework will help increase the application of machine-learning-based modeling approaches in hydrological sciences.
47
Huang, Yefei, Tianlai Xu, Zexu Zhang, Hutao Cui i Yu Su. "Satellite Segmentation with Pre-trained CNN Models". Journal of Physics: Conference Series 2171, nr 1 (1.01.2022): 012003. http://dx.doi.org/10.1088/1742-6596/2171/1/012003.
Pełny tekst źródłaStreszczenie:
Abstract In a generic satellite relative pose estimation pipeline, finding sufficient features in objects is quite essential to build the correct matching relationship and then solve the relative movement. However, for low-earth-orbit (LEO) satellites, since the earth background contains much more texture than objects, an object segmentation process is necessary to provide a prior range for feature extraction. In this work, we address this task with the pre-trained Deeplabv3 and fully convolutional network (FCN). Unlike the fine-tuning or transfer learning processes in other researches, we obtain probabilistic maps from the high-dimensional output of the above-mentioned CNN models and achieve a rough satellite extraction. Our method makes Deeplabv3 and FCN models work in a totally unfamiliar LEO scene and still achieves 0.2927 and 0.2122 in average intersection over union (IoU) respectively.
48
Miyazawa, Kazuki, Yuta Kyuragi i Takayuki Nagai. "Simple and Effective Multimodal Learning Based on Pre-Trained Transformer Models". IEEE Access 10 (2022): 29821–33. http://dx.doi.org/10.1109/access.2022.3159346.
Pełny tekst źródła
49
Shoaib, Mohamed R., Mohamed R. Elshamy, Taha E. Taha, Adel S. El-Fishawy i Fathi E. Abd El-Samie. "Efficient Brain Tumor Detection Based on Deep Learning Models". Journal of Physics: Conference Series 2128, nr 1 (1.12.2021): 012012. http://dx.doi.org/10.1088/1742-6596/2128/1/012012.
Pełny tekst źródłaStreszczenie:
Abstract Brain tumor is an acute cancerous disease that results from abnormal and uncontrollable cell division. Brain tumors are classified via biopsy, which is not normally done before the brain ultimate surgery. Recent advances and improvements in deep learning technology helped the health industry in getting accurate disease diagnosis. In this paper, a Convolutional Neural Network (CNN) is adopted with image pre-processing to classify brain Magnetic Resonance (MR) images into four classes: glioma tumor, meningioma tumor, pituitary tumor and normal patients, is provided. We use a transfer learning model, a CNN-based model that is designed from scratch, a pre-trained inceptionresnetv2 model and a pre-trained inceptionv3 model. The performance of the four proposed models is tested using evaluation metrics including accuracy, sensitivity, specificity, precision, F1_score, Matthew’s correlation coefficient, error, kappa and false positive rate. The obtained results show that the two proposed models are very effective in achieving accuracies of 93.15% and 91.24% for the transfer learning model and BRAIN-TUMOR-net based on CNN, respectively. The inceptionresnetv2 model achieves an accuracy of 86.80% and the inceptionv3 model achieves an accuracy of 85.34%. Practical implementation of the proposed models is presented.
50
Dang, Cach N., María N. Moreno-García i Fernando De la Prieta. "Hybrid Deep Learning Models for Sentiment Analysis". Complexity 2021 (12.08.2021): 1–16. http://dx.doi.org/10.1155/2021/9986920.
Pełny tekst źródłaStreszczenie:
Sentiment analysis on public opinion expressed in social networks, such as Twitter or Facebook, has been developed into a wide range of applications, but there are still many challenges to be addressed. Hybrid techniques have shown to be potential models for reducing sentiment errors on increasingly complex training data. This paper aims to test the reliability of several hybrid techniques on various datasets of different domains. Our research questions are aimed at determining whether it is possible to produce hybrid models that outperform single models with different domains and types of datasets. Hybrid deep sentiment analysis learning models that combine long short-term memory (LSTM) networks, convolutional neural networks (CNN), and support vector machines (SVM) are built and tested on eight textual tweets and review datasets of different domains. The hybrid models are compared against three single models, SVM, LSTM, and CNN. Both reliability and computation time were considered in the evaluation of each technique. The hybrid models increased the accuracy for sentiment analysis compared with single models on all types of datasets, especially the combination of deep learning models with SVM. The reliability of the latter was significantly higher.