Search

Physical neural networks (PNN) using physical materials and devices to mimic synapses and neurons offer an energy-efficient way to implement artificial neural networks. Yet, training PNN is difficult and heavily relies on external computing resources. An emerging concept to solve this issue is called physical self-learning that uses intrinsic physical parameters as trainable weights. Under external inputs (i.e., training data), training is achieved by the natural evolution of physical parameters that intrinsically adapt modern learning rules via an autonomous physical process, eliminating the requirements on external computation resources. Here, we demonstrate a real spintronic system that mimics Hopfield neural networks (HNN), and unsupervised learning is intrinsically performed via the evolution of the physical process. Using magnetic texture-defined conductance matrix as trainable weights, we illustrate that under external voltage inputs, the conductance matrix naturally evolves and adapts Oja's learning algorithm in a gradient descent manner. The self-learning HNN is scalable and can achieve associative memories on patterns with high similarities. The fast spin dynamics and reconfigurability of magnetic textures offer an advantageous platform toward efficient autonomous training directly in materials. [ABSTRACT FROM AUTHOR]

Optimization problems are common across various fields, and one effective solution is the swarm intelligence algorithm.It is essential for the algorithm to deliver high-quality solutions for problems with varying characteristics. However, most existing swarm intelligence rely on fixed and monotonic search strategies, which limits their ability to handle the diverse and complex situations encountered when solving real-world optimization problems with unknown fitness landscapes. To extend the applicability of swarm intelligence and thus offer users an efficient black-box optimizer for various applications, a novel self-learning mechanism is proposed and applied to the Salp Swarm Algorithm (SSA) to develop the self-learning salp swarm algorithm (SLSSA) in this paper. In SLSSA, four distinct search strategies, including a novel multiple food sources search strategy, are adopted to strengthen the search agents' abilities to conquer various difficulties in the search space. To improve the efficiency of the search process, the self-learning strategy dynamically determines the execution probability of each search strategy according to the quality of solutions it produced previously. Moreover, a parameter setting method is proposed in this paper, which eliminates the need for a trial-and-error approach and allows for straightforward configuration of the parameters that optimize the performance of SLSSA. In comparison with several highly regarded state-of-the-art peer algorithms, the performance of SLSSA in solving the CEC2014 benchmark functions was thoroughly examined. Subsequently, SLSSA was applied to train multi-layer perceptron classifiers and test on the UCI machine-learning datasets. The experimental results and analysis on benchmark functions and multi-layer perceptron classifier training problems demonstrate that SLSSA outperforms the competing algorithms in terms of solution accuracy, stability, and overall convergence speed. Moreover, computational time comparisons reveal that SLSSA achieves significant performance improvement with only a marginal increase in time cost compared to the original SSA. [ABSTRACT FROM AUTHOR]

Text annotation is an essential element of the natural language processing approaches. The manual annotation process performed by humans has various drawbacks, such as subjectivity, slowness, fatigue, and possibly carelessness. In addition, annotators may annotate ambiguous data. Therefore, we have developed the concept of automated annotation to get the best annotations using several machine-learning approaches. The proposed approach is based on an ensemble algorithm of meta-learners and meta-vectorizer techniques. The approach employs a semi-supervised learning technique for automated annotation to detect hate speech. This involves leveraging various machine learning algorithms, including Support Vector Machine (SVM), Decision Tree (DT), K-Nearest Neighbors (KNN), and Naive Bayes (NB), in conjunction with Word2Vec and TF-IDF text extraction methods. The annotation process is performed using 13,169 Indonesian YouTube comments data. The proposed model used a Stemming approach using data from Sastrawi and new data of 2245 words. Semi-supervised learning uses 5%, 10%, and 20% of labeled data compared to performing labeling based on 80% of the datasets. In semi-supervised learning, the model learns from the labeled data, which provides explicit information, and the unlabeled data, which offers implicit insights. This hybrid approach enables the model to generalize and make informed predictions even when limited labeled data is available (based on self-learning). Ultimately, this enhances its ability to handle real-world scenarios with scarce annotated information. In addition, the proposed method uses a variety of thresholds for matching words labeled with hate speech ranging from 0.6, 0.7, 0.8, to 0.9. The experiments indicated that the DT-TF-IDF model has the best accuracy value of 97.1% with a scenario of 5%:80%:0.9. However, several other methods have accuracy above 90%, such as SVM (TF-IDF and Word2Vec) and KNN (Word2Vec), based on both text extraction methods in several test scenarios. [ABSTRACT FROM AUTHOR]

Copyright of Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji is the property of Gazi University and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Complex and high-computational-cost algorithms are usually the state-of-the-art solution for autonomous driving cases in which non-holonomic robots must be controlled in scenarios with spatial restrictions and interaction with dynamic obstacles while fulfilling at all times safety, comfort, and legal requirements. These highly complex software solutions must cover the high variability of use cases that might appear in traffic conditions, especially when involving scenarios with dynamic obstacles. Reinforcement learning algorithms are seen as a powerful tool in autonomous driving scenarios since the complexity of the algorithm is automatically learned by trial and error with the help of simple reward functions. This paper proposes a methodology to properly define simple reward functions and come up automatically with a complex and successful autonomous driving policy. The proposed methodology has no motion planning module so that the computational power can be limited like in the reactive robotic paradigm. Reactions are learned based on the maximization of the cumulative reward obtained during the learning process. Since the motion is based on the cumulative reward, the proposed algorithm is not bound to any embedded model of the robot and is not being affected by uncertainties of these models or estimators, making it possible to generate trajectories with the consideration of non-holonomic constrains. This paper explains the proposed methodology and discusses the setup of experiments and the results for the validation of the methodology in scenarios with dynamic obstacles. A comparison between the reinforcement learning algorithm and state-of-the-art approaches is also carried out to highlight how the methodology proposed outperforms state-of-the-art algorithms. [ABSTRACT FROM AUTHOR]

This paper aims to develop an effective sensor fusion model for turning processes for the detection of tool wear. Fusion of sensors' data combined with novelty detection algorithm and learning vector quantisation (LVQ) neural networks is used to detect tool wear and present diagnostic and prognostic information. To reduce the number of sensors required in the monitoring system and support sensor fusion, the ASPS approach (Automated Sensor and Signal Processing Selection System) is used to select the most appropriate sensors and signal processing methods for the design of the condition monitoring system. The experimental results show that the proposed approach has demonstrated its efficacy in the implementation of an effective solution for the monitoring tool wear in turning. The results prove that the fusion of sensitive sensory characteristic features and the use of AI methods have been successful for the detection and prediction of the tool wear in turning processes and show the capability of the proposed approach to reduce the complexity of the design of condition monitoring systems and the development of a sensor fusion system using a self-learning method. [ABSTRACT FROM AUTHOR]

Modern manufacturing has to cope with dynamic and changing circumstances. Market fluctuations, the effects caused by unpredictable material shortages, highly variable product demand, and worker availability all require system robustness, flexibility, and resilience. To adapt to these new requirements, manufacturers should consider investigating, investing in, and implementing system autonomy. Autonomy is being adopted in multiple industrial contexts, but divergences arise when formalizing the concept of autonomous systems. To develop an implementation of autonomous manufacturing systems, it is essential to specify what autonomy means, how autonomous manufacturing systems are different from other autonomous systems, and how autonomous manufacturing systems are identified and achieved through the main features and enabling technologies. With a comprehensive literature review, this paper provides a definition of autonomy in the manufacturing context, infers the features of autonomy from different engineering domains, and presents a five-level model of autonomy — associated with maturity levels for the features — to ensure the complete identification and evaluation of autonomous manufacturing systems. The paper also presents the evaluation of a real autonomous system that serves as a use-case and a validation of the model. [ABSTRACT FROM AUTHOR]

The dramatic growth of smart devices and the Internet of things (IoT) has increased the volume of exchanges and data on the web. The centralized and traditional architecture of cloud computing does not meet the demands of users and proper implementation of latency-sensitive applications due to latency and mass demands of IoT applications that have different needs compared to existing applications. As a result, edge computing has been presented for collecting and processing of data generated by these objects, which facilitates data processing with low latency and close to users at the edge of the network. Its main purpose is to bring computational resources and storage close to the end-user on the network. As far as the storage capacity of fog nodes is limited, the proper use of fog node resources significantly influences their performance. In this paper, a framework based on control MAPE-K loop has been used to obtain the optimal state in workload balance. The users' workload forecasting model is a combination of linear regression and support vector regression methods and offers better performance compared to the conventional reactive self-assessment methods. In the planning phase, a fuzzy self-learning algorithm is used to determine the automated scale of resource provisioning policy. By comparing three criteria of load delay, cost, and amount of consumed energy in the proposed method and recent works, the proposed method has been able to balance all three criteria optimally. [ABSTRACT FROM AUTHOR]

Text annotation is an essential element of the natural language processing approaches. The manual annotation process performed by humans has various drawbacks, such as subjectivity, slowness, fatigue, and possibly carelessness. In addition, annotators may annotate ambiguous data. Therefore, we have developed the concept of automated annotation to get the best annotations using several machine-learning approaches. The proposed approach is based on an ensemble algorithm of meta-learners and meta-vectorizer techniques. The approach employs a semi-supervised learning technique for automated annotation to detect hate speech. This involves leveraging various machine learning algorithms, including Support Vector Machine (SVM), Decision Tree (DT), K-Nearest Neighbors (KNN), and Naive Bayes (NB), in conjunction with Word2Vec and TF-IDF text extraction methods. The annotation process is performed using 13,169 Indonesian YouTube comments data. The proposed model used a Stemming approach using data from Sastrawi and new data of 2245 words. Semi-supervised learning uses 5%, 10%, and 20% of labeled data compared to performing labeling based on 80% of the datasets. In semi-supervised learning, the model learns from the labeled data, which provides explicit information, and the unlabeled data, which offers implicit insights. This hybrid approach enables the model to generalize and make informed predictions even when limited labeled data is available (based on self-learning). Ultimately, this enhances its ability to handle real-world scenarios with scarce annotated information. In addition, the proposed method uses a variety of thresholds for matching words labeled with hate speech ranging from 0.6, 0.7, 0.8, to 0.9. The experiments indicated that the DT-TF-IDF model has the best accuracy value of 97.1% with a scenario of 5%:80%:0.9. However, several other methods have accuracy above 90%, such as SVM (TF-IDF and Word2Vec) and KNN (Word2Vec), based on both text extraction methods in several test scenarios.

Machine learning and specifically deep learning techniques address many of the issues faced in visual object detection and classification tasks. However, they have the caveat of needing large amounts of annotated training data. In the maritime domain one may encounter objects fairly infrequently, depending on weather and location. This creates an issue of data collection. Areas such as harbors and channels see a lot of traffic, but the ships are of a specific class. Furthermore, the variability of the buoys from region to region and within regions is difficult and expensive to sample. Thus the amount and quality of available data is severely lacking. Furthermore very few publicly available maritime datasets exist In this work, we present a novel approach that detects possible “poor” training samples and automatically re-annotates them, based on the current state of the object detector. We show the applicability of our approach on real-life maritime data and show that the poor annotation quality of the datasets used can be mitigated. We show performance gain with respect to a baseline approach is proportional to the amount of poorly annotated data in the dataset. When 25% of the data is poor we achieve a 5.5%, 13.7%, and 8.0% increase in performance on 3 separate datasets, compared to a baseline model. With 50% noise we reach 58.5%, 18.7% and 94.2% increase respectively. Our approach also allows for the iterative improvement of a given dataset by providing a set of pseudo-annotations to replace the current incorrect ones.

Smart home systems are the integration of technology and services through the network for a better quality of life. Smart homes perform daily housework and activities more easily without user intervention or with remote control of the user. In this study, a machine learning-based smart home system has been developed. The aim of the study is to design a system that can continuously improve itself and learn instead of an ordinary smart home system that can be remotely controlled with the help of machine learning. The developed machine learning model predicts the routine activities of the users in the home and performs some operations for the user autonomously. The dataset used in the study consists of real data received from the sensors as a result of the daily use. Naive Bayes (NB) (i.e. Gaussian NB, Bernoulli NB, Multinomial NB, and Complement NB), ensemble (i.e. Random Forest, Gradient Tree Boosting and eXtreme Gradient Boosting), linear (i.e. Logistic Regression, Stochastic Gradient Descent, and Passive-Aggressive Classification), and other (i.e. Decision Tree, Support Vector Machine, K Nearest Neighbor, Gaussian Process Classifier (GPC), Multilayer Perceptron) machine learning-based algorithms were utilized. The performance of the proposed smart home system was evaluated using several performance metrics: The best results were obtained from the GPC algorithm (i.e. Precision: 0.97, Recall: 0.98, F1-score: 0.97, Accuracy: 0.97). [ABSTRACT FROM AUTHOR]

In recent years, the recommendation system and robot learning are undoubtedly the two most popular application fields, and the core algorithms supporting these two fields are deep learning based on perception and reinforcement learning based on exploration learning, respectively. How to combine these two fields to better improve the development of the whole machine learning field is the dream of numerous researchers. The Deep Reinforcement Network (DRN) model successfully embedded reinforcement learning into the recommendation system, which provided a good idea for subsequent researchers. However, the disadvantage is also obvious, that is, the DRN model is built for news recommendations, meaning that the DRN model is not transferable, which is also the defect of many current recommendation system models. Meanwhile, the agent learning method adopted by the DRN model is primitive and inefficient. Among many models and algorithms that have emerged in recent years, we use the newly proposed deployment efficiency to measure their comprehensive quality and found that few models focus on both efficiency and performance improvement. To fill the gap of model deployment efficiency neglected by many researchers and to create a model of reinforcement learning agents with stronger performance, we have been exploring and trying to complete research on the Gate Attentional Factorization Machines (GAFM) model. Finally, we successfully integrated the GAFM model and reinforcement learning. The Deep Reinforcement Factorization Machines (DRFM) model proposed in this paper is based on the combination of deep learning with high perception ability and reinforcement learning with high exploration ability, centered on improving the deployment efficiency and learning performance of the model. The GAFM model is modified and upgraded using multidisciplinary techniques, and a new model-based random exploration strategy is proposed to update and optimize the recommendation list efficiently. Through parallel contrast experiments on various datasets, it is proved that the DRFM model surpasses the traditional recommendation system model in all aspects. The DRFM model is far superior to other models in terms of performance and robustness, and also significantly improved in terms of deployment efficiency. At the same time, we conduct a comparative analysis with the latest deep reinforcement learning algorithm and prove the unique advantages of the DRFM model. [ABSTRACT FROM AUTHOR]

In recent years, personalized dialogue generation has attracted researchers' attention due to its wide range of applications. Although there has been much excellent research, current chatbots are known to generate uninteresting responses without personality. In a general way, persona information is adopted to mitigate these issues. Yet, how to integrate persona information reasonably based on historical dialogue in responses is still a severe challenge. Thus, we present a solution that consolidates the persona information memory process and pretrained method to generate personalized and fluency responses. Moreover, we adopt a reinforcement learning algorithm to joint each part of the dialogue model. Last but not least, we present a self-learning framework to explore the hypothetical space, and make the responses more personalized and fascinating. Extensive experiments on the large-scale dialogue public dataset ConvAI2 verify the effectiveness of our method. [ABSTRACT FROM AUTHOR]

Named entity recognition (NER) constitutes an important step in the processing of unstructured text content for the extraction of information as well as for the computer-supported analysis of large amounts of digital data via machine learning methods. However, NER often relies on domain-specific knowledge, being conducted manually in a time- and human-resource-intensive process. These can be reduced with statistical models performing NER automatically. The current work investigates whether Conditional Random Fields (CRF) can be efficiently trained for NER in German texts, by means of an iterative procedure combining self-learning with a manual annotation–active learning–component. The training dataset increases continuously with the iterative procedure. Whilst self-learning did not markedly improve the performance of the CRF for NER, the manual annotation of sentences with the lowest probability of correct prediction clearly improved the model F1-score and simultaneously reduced the amount of manual annotation required to train the model. A model with an F1-score of 0.885 was able to be trained in 11.4 h. [ABSTRACT FROM AUTHOR]

Copyright of Dokuz Eylul University Muhendislik Faculty of Engineering Journal of Science & Engineering / Dokuz Eylül Üniversitesi Mühendislik Fakültesi Fen ve Mühendislik Dergisi is the property of Dokuz Eylul Universitesi Muhendislik Fakultesi Fen ve Muhendislik Dergisi and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The trend of bringing machine learning (ML) to the Internet of Things (IoT) field devices is becoming ever more relevant, also reducing the overall energy need of the applications. ML models are usually trained in the cloud and then deployed on edge devices. Most IoT devices generate large amounts of unlabeled data, which are expensive and challenging to annotate. This paper introduces the self-learning autonomous edge learning and inferencing pipeline (AEP), deployable in a resource-constrained embedded system, which can be used for unsupervised local training and classification. AEP uses two complementary approaches: pseudo-label generation with a confidence measure using k-means clustering and periodic training of one of the supported classifiers, namely decision tree (DT) and k-nearest neighbor (k-NN), exploiting the pseudo-labels. We tested the proposed system on two IoT datasets. The AEP, running on the STM NUCLEO-H743ZI2 microcontroller, achieves comparable accuracy levels as same-type models trained on actual labels. The paper makes an in-depth performance analysis of the system, particularly addressing the limited memory footprint of embedded devices and the need to support remote training robustness.

In modern wood processing enterprises, the main quality classification factors of solid wood lumbers are defects and textures. This research proposed a deep learning algorithm based on a mixture of local binary pattern, self-learning deep belief networks and softmax classification to classify solid wood lumbers with defects and textures. Firstly, it extracted the defects and textures features of solid wood lumbers, on which to be based, used the deep belief networks to train and learn the characteris tics of the local binarized processing. Afterwards, it adopted the self-learning learning rate algorithm to optimize the convergence speed and reduced the training time. Finally, the algorithm obtained the common defects, straight textures and confused textures by using a softmax classification. Compared with classical algorithms e. g. BP neural network, support vector machine and extreme learning machine, the error rate of solid wood defects and textures recognition obtained by the deep belief networks designed is about 3. 59%, and this algorithm works efficiently in recognition the defects and textures of solid wood lumbers. [ABSTRACT FROM AUTHOR]

A huge amount of data is generated since the evolution in technology and the tremendous growth of social networks. In spite of the availability of data, there is a lack of tools and resources for analysis. Though Arabic is a popular language, there are too few dialectal Arabic analysis tools. This is because of the many challenges in Arabic due to its morphological complexity and dynamic nature. Sentiment Analysis (SA) is used by different organizations for many reasons as developing product quality, adjusting market strategy and improving customer services. This paper introduces a semi- automatic learning system for sentiment analysis that is capable of updating the lexicon in order to be up to date with language changes. HILATSA is a hybrid approach which combines both lexicon based and machine learning approaches in order to identify the tweets sentiments polarities. The proposed approach has been tested using different datasets. It achieved an accuracy of 73.67% for 3-class classification problem and 83.73% for 2-class classification problem. The semi-automatic learning component proved to be effective as it improved the accuracy by 17.55%. [ABSTRACT FROM AUTHOR]

Self-learning super-resolution (SLSR) algorithms have the advantage of being independent of an external training database. This paper proposes an SLSR algorithm that uses convolutional principal component analysis (CPCA) and random matching. The technologies of CPCA and random matching greatly improve the efficiency of self-learning. There are two main steps in this algorithm: forming the training and testing the data sets and patch matching. In the data set forming step, we propose the CPCA to extract the low-dimensional features of the data set. The CPCA uses a convolutional method to quickly extract the principal component analysis (PCA) features of each image patch in every training and testing image. In the patch matching step, we propose a two-step random oscillation accompanied with propagation to accelerate the matching process. This patch matching method avoids exhaustive searching by utilizing the local similarity prior of natural images. The two-step random oscillation first performs a coarse patch matching using the variance feature and then performs a detailed matching using the PCA feature, which is useful to find reliable matching patches. The propagation strategy enables patches to propagate the good matching patches to their neighbors. The experimental results demonstrate that the proposed algorithm has a substantially lower time cost than that of many existing self-learning algorithms, leading to better reconstruction quality. [ABSTRACT FROM AUTHOR]

Entity alignment aims to identify semantical matchings between entities from different groups. Traditional methods (e.g., attribute comparison-based methods, graph operation-based methods and active learning ones) are usually supervised by labeled data as prior knowledge. Since it is not trivial to label data for training, researchers have then turned to unsupervised methods, and have thus developed similarity-based methods, probabilistic methods, graphical model-based methods, etc. In addition, structure or class information is further explored. As an important part of a knowledge graph, entities contain rich semantical information that can be well learned by knowledge graph embedding methods in low-dimensional vector spaces. However, existing methods for entity alignment have paid little attention to knowledge graph embedding. In this paper, we propose a self-learning and embedding based method for entity alignment, thus called SEEA, to iteratively find semantically aligned entity pairs, which makes full use of semantical information contained in the attributes of entities. Experiments on three realistic datasets and comparison with a few baseline methods validate the effectiveness and merits of the proposed method. [ABSTRACT FROM AUTHOR]

An improved teaching-learning-based optimization (I-TLBO) algorithm is proposed to adjust the parameters of extreme learning machine with parallel layer perception (PELM), and a well-generalized I-TLBO-PELM model is obtained to build the model of NOX emissions of a boiler. In the I-TLBO algorithm, there are four major highlights. Firstly, a quantum initialized population by using the qubits on Bloch sphere replaces a randomly initialized population. Secondly, two kinds of angles in Bloch sphere are generated by using cube chaos mapping. Thirdly, an adaptive control parameter is added into the teacher phase to speed up the convergent speed. And then, according to actual teaching-learning phenomenon of a classroom, students learn some knowledge not only by their teacher and classmates, but also by themselves. Therefore, a self-study strategy by using Gauss mutation is introduced after the learning phase to improve the exploration ability. Finally, we test the performance of the I-TLBO-PELM model. The experiment results show that the proposed model has better regression precision and generalization ability than eight other models. [ABSTRACT FROM AUTHOR]

In this paper, we proposed the Dandelion Algorithm (DA), based on the behaviour of dandelion sowing. In DA, the dandelion is sown in a certain range based on dynamic radius. Meanwhile the dandelion has self-learning ability; it could select a number of excellent seeds to learn. We compare the proposed algorithm with other existing algorithms. Simulations show that the proposed algorithm seems much superior to other algorithms. Moreover, the proposed algorithm can be applied to optimise extreme learning machine (ELM), which has a very good classification and prediction capability. [ABSTRACT FROM PUBLISHER]

Sketch recognition is the task of converting hand-drawn digital ink into symbolic computer representations. Since the early days of sketch recognition, the bulk of the work in the field focused on building accurate recognition algorithms for specific domains, and well defined data sets. Recognition methods explored so far have been developed and evaluated using standard machine learning pipelines and have consequently been built over many simplifying assumptions. For example, existing frameworks assume the presence of a fixed set of symbol classes, and the availability of plenty of annotated examples. However, in practice, these assumptions do not hold. In reality, the designer of a sketch recognition system starts with no labeled data at all, and faces the burden of data annotation. In this work, we propose to alleviate the burden of annotation by building systems that can learn from very few labeled examples, and large amounts of unlabeled data. Our systems perform self-learning by automatically extending a very small set of labeled examples with new examples extracted from unlabeled sketches. The end result is a sufficiently large set of labeled training data, which can subsequently be used to train classifiers. We present four self-learning methods with varying levels of implementation difficulty and runtime complexities. One of these methods leverages contextual co-occurrence patterns to build verifiably more diverse set of training instances. Rigorous experiments with large sets of data demonstrate that this novel approach based on exploiting contextual information leads to significant leaps in recognition performance. As a side contribution, we also demonstrate the utility of bagging for sketch recognition in imbalanced data sets with few positive examples and many outliers. [ABSTRACT FROM AUTHOR]

Precise carbon price forecasting can provide decision support for policy-makers and investors. However, due to the high non-stationarity and nonlinearity of carbon price series, it is difficult to get accurate forecasting results under volatile situations. To accommodate different scenarios, this paper proposes a dynamic self-learning integrating forecasting model to forecast the carbon price by considering external impact factors. The multi-dimensional time series is initially decomposed into different intrinsic mode functions simultaneously by the noise-assisted multivariate empirical mode decomposition method. After reconstructing the decomposed series into high-frequency, low-frequency, and trend modules, the extreme learning machine optimized by the cosine-based whale optimization algorithm is proposed to predict the carbon price. The dynamic relationships between the carbon price and impact factors are simulated by the sliding window structure, which improves the adaptability of the proposed model. The high prediction accuracy under different situations including extreme scenarios demonstrates the stability of the proposed model. A self-learning algorithm, which can automatically learn the evolving model structure and update model parameters, is designed to alleviate the underfitting/overfitting problem. The comparison results with existing models indicate the superiority of the proposed model. • The dynamic self-learning N-A MEMD-COSWOA-ELM model gets accurate predictions. • The WOA has been improved to be COSWOA by emphasizing the global search. • The sliding window structure catches the time-varying characteristics of the multivariate series. • A self-learning algorithm is designed to address the under-/over-fitting problem. [ABSTRACT FROM AUTHOR]

The emergence of new horizons in the field of travel assistant management leads to the development of cutting-edge systems focused on improving the existing ones. Moreover, new opportunities are being also presented since systems trend to be more reliable and autonomous. In this paper, a self-learning embedded system for object identification based on adaptive-cooperative dynamic approaches is presented for intelligent sensor's infrastructures. The proposed system is able to detect and identify moving objects using a dynamic decision tree. Consequently, it combines machine learning algorithms and cooperative strategies in order to make the system more adaptive to changing environments. Therefore, the proposed system may be very useful for many applications like shadow tolls since several types of vehicles may be distinguished, parking optimization systems, improved traffic conditions systems, etc. [ABSTRACT FROM AUTHOR]

Abstract: The next big thing in computing is the Internet of Things (IoT), referring to dynamic and ever-evolving environments, generating high-volume streams of heterogeneous yet correlated contextual information of varying quality and complexity. Moreover, the increase in user mobility and unreliable sensor availability in IoT, necessitates the context-aware applications to dynamically adapt their behavior at run time. In this paper, we elicit the need for different self-learning techniques to tackle the openness of the IoT environments and propose enabling algorithms to achieve them. First, we present possible application scenarios which can benefit from both supervised and unsupervised self-learning. Later, we propose correlation mining algorithms based on Kullback-Leibler (KL) divergence and frequent set mining that exploits correlated contexts to enable unsupervised self-learning. These algorithms help to identify alternate sources for a context and semantically describe the previously unseen contexts in terms of already known contexts. We have realized the proposed algorithms on top of a Bayesian framework (HARD-BN) which supports autonomous learning. Our experiments demonstrate the applicability of the proposed correlation mining algorithms and their feasibility to enable self-learning in open and ever-evolving IoT environments. [Copyright &y& Elsevier]

Highlights: [•] The architecture of a self-learning, ontology-based framework is presented. [•] Framework enables context-aware applications to change their behavior at run-time. [•] Knowledge added to the ontology is given a probability to convey its reliability. [•] An implemented use case is presented: mining the reasons for patients’ nurse calls. [•] Datasets of size >1000 with <5% noise yield correct results in negligible time. [Copyright &y& Elsevier]

Recently, the application of the maintenance transmission line robot has been very popular in the power industry. However, difficulties in the control of maintenance transmission line robot exist due to multiple nonlinearities, plant parameter variations and external disturbances. This paper investigates the possibility of using neural network as a promising self-learning control alternative for the control problem of inspection and deicing transmission line robot. We first discuss the mechanical structure, as well as dynamic model of a deicing robot. And then, a neural network-based self-learning control strategy consists of a fuzzy neural network controller and an ELM-based single-layer-feedback neural networks identifier are proposed for this deicing transmission line robot. Both the structure and the learning algorithm of the control system are presented. The proposed controller is verified by computer simulations and experiments. [ABSTRACT FROM AUTHOR]

We describe the design of an autonomous agent that can teach itself how to translate from a foreign language, by first assembling its own training set, then using it to improve its vocabulary and language model. The key idea is that a Statistical Machine Translation package can be used for the Cross-Language Retrieval Task of assembling a training set from a vast amount of available text (e.g. a large multilingual corpus, or the Web) and then train on that data, repeating the process several times. The stability issues related to such a feedback loop are addressed by a mathematical model, connecting statistical and control-theoretic aspects of the system. We test it on controlled environment and real-world tasks, showing that indeed this agent can improve its translation performance autonomously and in a stable fashion, when seeded with a very small initial training set. We develop a multiprocessor version of the agent that directly accesses the Web using a Web search engine and taking advantage of the big amount of data available there. The modelling approach we develop for this agent is general, and we believe that it will be useful for an entire class of self-learning autonomous agents working on the Web. [ABSTRACT FROM AUTHOR]

Abstract: A self-learning method in intelligent instruction system is presented in this paper. Some concepts of information theory are introduced firstly. Then we analyze the machine learning mechanism and propose a self-learning arithmetic to reduce the magnitude of the decision tree. In the end this arithmetic is applied to intelligent instruction system. [Copyright &y& Elsevier]

[Display omitted] • A unified local search framework is proposed for feature selection wrappers. • Neighborhoods of the local search effect the exploration and exploitation. • Feature attributions can improve the efficiency to search for optimal subsets. Feature selection has shown its effectiveness in improving the accuracy and generalization of machine learning models, especially for those tasks with high-dimensional data. In this article, a novel self-learning feature selection (SLFS) approach based on feature attributions is proposed as a wrapper method, which has higher search efficiency for optimal feature subsets with three main improvements. First, we regard feature selection as a combinatorial optimization problem and propose a unified local search framework for wrapper methods by analyzing meta-heuristic algorithms in feature selection. Second, for the binary search space of feature selection, we propose two types of neighborhood structures, namely, ring-type and line-type structures, for the local search framework. Third, we focus on feature attribution methods, such as SHAP (SHapley Additive exPlanations) (Lundberg & Lee, 2017), which can interpret each feature's importance to predictions. In each iteration, we adopt SHAP values and other attributes from previous subsets to guide the next selection of new subsets. To validate the performance of our SLFS approach, we collected 16 classification datasets from the UCI repository for comparison with other meta-heuristic wrapper approaches in terms of fitness, accuracy, F1 scores and selection ratios. The experimental results show that the SLFS approach can be used to obtain an optimal subset with fewer iterations and a small population, and SHAP values play a role in improving search efficiency. [ABSTRACT FROM AUTHOR]

The trend of bringing machine learning (ML) to the Internet of Things (IoT) field devices is becoming ever more relevant, also reducing the overall energy need of the applications. ML models are usually trained in the cloud and then deployed on edge devices. Most IoT devices generate large amounts of unlabeled data, which are expensive and challenging to annotate. This paper introduces the self-learning autonomous edge learning and inferencing pipeline (AEP), deployable in a resource-constrained embedded system, which can be used for unsupervised local training and classification. AEP uses two complementary approaches: pseudo-label generation with a confidence measure using k-means clustering and periodic training of one of the supported classifiers, namely decision tree (DT) and k-nearest neighbor (k-NN), exploiting the pseudo-labels. We tested the proposed system on two IoT datasets. The AEP, running on the STM NUCLEO-H743ZI2 microcontroller, achieves comparable accuracy levels as same-type models trained on actual labels. The paper makes an in-depth performance analysis of the system, particularly addressing the limited memory footprint of embedded devices and the need to support remote training robustness. [ABSTRACT FROM AUTHOR]