Search

The complex traffic and road environment pose considerable challenges to the accuracy, timeliness, and adaptive ability of connected and autonomous vehicles (CAVs) in making driving decisions. This paper uses vehicle collaboration and integrates the adaptive learning capabilities of machine learning and the interpretation capabilities of expert systems (ESs) in a unified architecture to form a hybrid autonomous driving guidance system, which not only solves the ``bottleneck'' of knowledge acquisition during the construction of expert systems but also solves the ``black box'' phenomenon of machine learning in the decision-making process. First, an autonomous driving strategy based on deep reinforcement learning (DRL) is proposed for CAVs to make decisions and extract corresponding rules. Next, we design an ES knowledge base expansion method including rule extraction, rule sharing, and rule test. Particularly, vehicular blockchain is adopted to ensure user privacy and data security during the rule-sharing process. Third, hybrid autonomous driving guidance combining ES and machine learning is proposed for CAVs to make accurate and efficient decisions in different driving environments. Once the strategy is well trained, it can effectively guide CAVs to cope with the complex traffic environment. Extensive simulations validate the performance of our proposal in terms of decision-making accuracy, effectiveness, and safety.

Improving the detection accuracy of pulmonary nodules plays an important role in the diagnosis and early treatment of lung cancer. In this paper, a multiscale aggregation network (MSANet), which integrates spatial and channel information, is proposed for 3D pulmonary nodule detection. MSANet is designed to improve the network's ability to extract information and realize multiscale information fusion. First, multiscale aggregation interaction strategies are used to extract multilevel features and avoid feature fusion interference caused by large resolution differences. These strategies can effectively integrate the contextual information of adjacent resolutions and help to detect different sized nodules. Second, the feature extraction module is designed for efficient channel attention and self-calibrated convolutions (ECA-SC) to enhance the interchannel and local spatial information. ECA-SC also recalibrates the features in the feature extraction process, which can realize adaptive learning of feature weights and enhance the information extraction ability of features. Third, the distribution ranking (DR) loss is introduced as the classification loss function to solve the problem of imbalanced data between positive and negative samples. The proposed MSANet is comprehensively compared with other pulmonary nodule detection networks on the LUNA16 dataset, and a CPM score of 0.920 is obtained. The results show that the sensitivity for detecting pulmonary nodules is improved and that the average number of false-positives is effectively reduced. The proposed method has advantages in pulmonary nodule detection and can effectively assist radiologists in pulmonary nodule detection.

Deep learning techniques have recently brought many improvements in the field of neural network training, especially for prognosis and health management. The success of such an intelligent health assessment model depends not only on the availability of labeled historical data but also on the careful samples selection. However, in real operating systems such as induction machines, which generally have a long reliable life, storing the entire operation history, including deterioration (i.e. bearings), will be very expensive and difficult to feed accurately into the training model. Other alternatives sequentially store samples that hold degradation patterns similar to real ones in damage behavior by imposing an accelerated deterioration. Labels lack and differences in distributions caused by the imposed deterioration will ultimately discriminate the training model and limit its knowledge capacity. In an attempt to overcome these drawbacks, a novel sequence-by-sequence deep learning algorithm able to expand the generalization capacity by transferring obtained knowledge from life cycles of similar systems is proposed. The new algorithm aims to determine health status by involving long short-term memory neural network as a primary component of adaptive learning to extract both health stage and health index inferences. Experimental validations are performed using the PRONOSTIA bearing degradation datasets.

In this article, an adaptive learning framework is established for a deep weighted sparse autoencoder (AE) by resorting to the multiobjective evolutionary algorithm (MOEA). The weighted sparsity is introduced to facilitate the design of the varying degrees of the sparsity constraints imposed on the hidden units of the AE. The MOEA is exploited to adaptively seek appropriate hyperparameters, where the divide-and-conquer strategy is implemented to enhance the MOEA's performance in the context of deep neural networks. Moreover, a sharing scheme is proposed to further reduce the time complexity of the learning process at the slight expense of the learning precision. It is shown via extensive experiments that the established adaptive learning framework is effective, where different sparse models are utilized to demonstrate the generality of the proposed results. Then, the generality of the proposed framework is examined on the convolutional AE and VGG-16 network. Finally, the developed framework is applied to the blind image quantity assessment that illustrates the applicability of the established algorithms.

The predictive vessel surveillance is one of the indispensable functional components in intelligent maritime traffic system. Vessel trajectory prediction serves as a prerequisite for collision detection and risk assessment. Perceiving the forthcoming traffic situation in advance helps decide the succeeding actions to mitigate the potential risk. The availability of maritime big data brings great potential to extract vessel movement patterns to support trajectory forecasting. In this paper, a novel vessel trajectory and navigating state prediction methodology is proposed based on AIS data, which synergizes properly designed learning, motion modelling and knowledge base assisted particle filtering processes. The primary contributions of this work also comprise several critical research findings to handle the key challenges in vessel trajectory and navigating state prediction problem, such as the adaptive training window determination for the learning process and effective knowledge storage and searching algorithm intended to reduce the query time of waterway pattern retrieval. The studies for these challenges are still missing in the reported literatures but they are essentially important for improving the prediction accuracy, efficiency and practicality. With the maritime traffic data collected for Singapore water, a thorough evaluation of the prediction performance has been conducted for different navigating scenarios. It is also observed that better prediction outperforms on account of allowing earlier alert in risk detection.

Exploitability prediction has become increasingly important in cybersecurity, as the number of disclosed software vulnerabilities and exploits are soaring. Recently, machine learning and deep learning algorithms, including Support Vector Machine (SVM), Decision Tree, deep Neural Networks and their ensemble models, have achieved great success in vulnerability evaluation and exploitability prediction. However, they make a strong assumption that the data distribution is static over time and therefore fail to consider the concept drift problems due to the evolving system behaviours. In this work, we propose a novel consecutive batch learning algorithm, called Real-time Dynamic Concept Adaptive Learning (RDCAL), to deal with the concept drift and dynamic class imbalance problems existing in exploitability prediction. Specifically, we develop a Class Rectification Strategy (CRS) to handle the ‘actual drift’ in sample labels and a Balanced Window Strategy (BWS) to boost the minority class during real-time learning. Experimental results conducted on the real-world vulnerabilities collected between 1988 to 2020 show that the overall performance of classifiers, including Neural Networks, SVM, HoeffdingTree and Logistic Regression (LR), improves over 3% by adopting our proposed RDCAL algorithm. Furthermore, RDCAL achieves state-of-the-art performance on exploitability prediction compared with other concept drift algorithms.

POI recommendation has become an essential means to help people discover attractive places. Intuitively, activities have an important impact on users’ decision-making, because users select POIs to attend corresponding activities. However, many existing studies ignore the social motivation of user behaviors and regard all check-ins as influenced only by individual user interests. As a result, they cannot model user preferences accurately, which degrades recommendation effectiveness. In this article, from the perspective of activities, this study proposes a probabilistic generative model called STARec. Specifically, based on the social effect of activities, STARec defines users’ social preferences as distinct from their individual interests and combines these with individual user activity interests to effectively depict user preferences. Moreover, the inconsistency between users’ social preferences and their decisions is modeled. An activity frequency feature is introduced to acquire accurate user social preferences because of close correlation between these and the key impact factor of corresponding check-ins. An alias sampling-based training method was used to accelerate training. Extensive experiments were conducted on two real-world datasets. Experimental results demonstrated that the proposed STARec model achieves superior performance in terms of high recommendation accuracy, robustness to data sparsity, effectiveness in handling cold-start problems, efficiency, and interpretability.

Industrial prognosis refers to the prediction of failures of an industrial asset based on data collected by Internet of Things sensors. Prognostic models can experience the undesired effects of concept drift, namely, the presence of nonstationary phenomena that affects the data collected over time. Consequently, fault patterns learned from data become obsolete. To overcome this issue, contextual and operational changes must be detected and managed, triggering rapid model adaptation mechanisms. This article presents an adaptive learning approach based on a dendritic cell algorithm for drift detection and a deep neural network model that dynamically adapts to new operational conditions. A kernel density estimator with drift-based bandwidth is used to generate synthetic data for a faster adaptation, focusing on fine-tuning the lowest neural layers. Experimental results over a real-world industrial problem shed light on the outperforming behavior of the proposed approach when compared to other drift detectors and classification models.

The sudden and spontaneous occurrence of epileptic seizures can impose a significant burden on patients with epilepsy. If seizure onset can be prospectively predicted, it could greatly improve the life of patients with epilepsy and also open new therapeutic avenues for epilepsy treatment. However, discovering effective predictive patterns from massive brainwave signals is still a challenging problem. The prediction of epileptic seizures is still in its early stage. Most existing studies actually investigated the predictability of seizures offline instead of a truly prospective online prediction, and also the high inter-individual variability was not fully considered in prediction. In this study, we propose a novel adaptive pattern learning framework with a new online feature extraction approach to achieve personalized online prospective seizure prediction. In particular, a two-level online feature extraction approach is applied to monitor intracranial electroencephalogram (EEG) signals and construct a pattern library incrementally. Three prediction rules were developed and evaluated based on the continuously-updated patient-specific pattern library for each patient, including the adaptive probabilistic prediction (APP), adaptive linear-discriminant-analysis-based prediction (ALP), and adaptive Naive Bayes-based prediction (ANBP). The proposed online pattern learning and prediction system achieved impressive prediction results for 10 patients with epilepsy using long-term EEG recordings. The best testing prediction accuracy averaged over the 10 patients were 79, 78, and 82 percent for the APP, ALP, and ANBP prediction scheme, respectively. The experimental results confirmed that the proposed adaptive prediction framework offers a promising practical tool to solve the challenging seizure prediction problem.

A learning style that focuses on individual learning is one of the most important aspects of any learning environment. Each learner has a unique manner of understanding, retaining, processing, and interpreting new information based on their learning styles. The ability of an e-learning system to automatically determine a student's learning style has become more essential. For learning events, the evolution of e-learning platforms provides students with higher opportunities online. In this paper, we proposed a Convolutional Neural Network-based Levy Flight Distribution (CNN-LFD) algorithm for learning style prediction. An adaptive e-learning system is divided into two sections: automatic learning style prediction and classification based on the number of learning styles included. Initially, the student logs in with their user ID, and the data is saved in the database. The features such as questionnaire score, login credentials (session ID, learner ID, and course ID), and login time (location, session ID) are extracted along with the sequence of the user's learning behavior. After that, the CNN-LFD algorithm predicts the learning styles of the learners namely Active/reflective, Sensing/intuitive, visual/verbal, sequential/global based on the extracted features. The dataset information are gathered from a Massive Open Online Course (MOOC), and the proposed model is built in JAVA software. The experimental results demonstrate higher classification accuracy during learning style prediction. The proposed CNN-LFD algorithm accomplishes 97.09% accuracy, 94.76% specificity, 92.12% sensitivity, and 97.56%, precision values than other methods.

Metaheuristics are powerful and generic global search methods. Most metaheuristics methods are not fully equipped with learning processes. Therefore, most of the search history is not reused in further steps of metaheuristics. The main aim of this research is to develop a general framework for automating and enhancing the search process and procedures in metaheuristics. The proposed framework, called Global Sensing Search (GSS), utilizes search memories to equip the search with applicable sensing features and adaptive learning elements to find a better solution and explore more diverse ones. Moreover, the GSS framework applies different search conditions to check the need for using suitable intensification and/or diversification strategies and also for terminating the search. An implementation of the GSS framework is proposed to alter the structure of standard genetic algorithms (GAs). Therefore, a new GA-based method called Genetic Sensing Algorithm is presented. The computational experiments show the efficiency of the proposed methods.

With increasing research on distributed processing in networks, adaptive learning strategies have gradually attracted researchers’ attention. The traditional adaptive learning strategy mainly aims at a single task unchanged over time, while real networks often entail multitasks scenarios with tasks that change over time. Furthermore, although cooperation among agents is beneficial for single task time-invariant networks, agents’ indiscriminate cooperation in time-varying multitask networks may cause undesirable effects. In this paper, an adaptive clustering learning approach based on an event-triggered scheme over time-varying multitask networks is proposed. With this process, agents are enabled to distinguish their clusters to determine cooperative neighbors and improve the accuracy of estimation over networks. The mean-square of the proposed algorithm is analyzed in detail, and the error probability of false alarms and false detections of the clustering mechanism is evaluated. An extensive simulation is given to validate the analytical performance of the distributed learning strategy over time-varying multitasks.

A goal in ultrasonic welding (USW) process monitoring is to accurately predict quality outcomes based on monitored signals. However, in most cases, knowing only that the USW process has failed is insufficient. Modern process automation should assess signal information and intercede to rectify process problems. Identification of when a process signal deviates from an acceptable final quality outcome, i.e., the time at which an abnormal event starts, facilitates control action or root cause analysis to bring it back to compliance. A long short-term memory (LSTM) recurrent neural network is proposed to monitor USW and other time-series signals and identify this point. This deep neural network is trained to classify quality outcomes from continuous signals. The process monitoring signals and their sampling time are divided into finite segments as input to this network. The time segment at which the process signal first converges to the final quality class prediction is identified using cross-entropy of the classification probabilities. This procedure is demonstrated using USW quality monitoring algorithms and robot motion failure detection. The examples show an LSTM network not only provides high accuracy for USW quality prediction, but also that the time of classification convergence is consistent with variance observed in USW weld quality factors. Moreover, classification convergence time was shown to be associated to specific robot motion failures, useful as input to adaptive learning. This work realizes deep-learning driven quality prediction and early event detection for quality classification problems, and provides the information necessary for adaptive control algorithms.

e-Learning is becoming the learning system of the future. However, many recent researches suggested replacing the "one size fits all" teaching model with the smarter model of adaptive learning in which the interpersonal differences between learners are considered the main shapers of lesson for a particular student. However, the main hurdle for the adaptive model is building the repository of micro learning objects (MLOs) with reasonable sizes suitable for reassembling into lessons in a way that is more suitable for a particular student. The process of designing such MLOs is expensive and time-consuming, reasons with the least effect of dampening the model to the point that it may collapse. Noteworthy, there is a wealth of digital learning contents and open-source educational curricula available online, but in the large granular traditional format. Therefore, this research proposes a Text processing intelligent framework for automatically identifying, extracting, and, annotating, those MLOs out of open digital video contents and PowerPoint courses, which are the most popular media types. أصبح التعلم الإلکتروني نظام التعلم في المستقبل. ومع ذلک ، اقترحت العديد من الأبحاث الحديثة استبدال نموذج التدريس "مقاس واحد يناسب الجميع" بالنموذج الأکثر ذکاء للتعلم التکيفي الذي تعتبر فيه الفروق الشخصية بين المتعلمين أهم صانعي الدرس لطالب معين. ومع ذلک ، فإن العقبة الرئيسية للنموذج التکيفي هي بناء مستودع کائنات التعلم الجزئيMLOsبأحجام معقولة مناسبة لإعادة التجميع في الدروس بطريقة أکثر ملاءمة لطالب معين. إن عملية تصميم مثل هذه العمليات مکلفة وتستغرق وقتا طويلا ، وهي أسباب أقل تأثيرا لإخماد النموذج لدرجة أنه قد ينهار. من الجدير بالذکر أن هناک ثروة من محتويات التعلم الرقمي والمناهج التعليمية مفتوحة المصدر المتاحة عبر الإنترنت ، ولکن بتنسيق تقليدي محبب کبير. لذلک ، يقترح هذا البحث إطارا ذکيا لمعالجة النصوص لتحديد واستخراج وتعليق ، تلقائيا ، MLOs من محتويات الفيديو الرقمية المفتوحة ودورات PowerPoint ، وهي أکثر أنواع الوسائط شيوعا.

Analyzing healthcare data poses several challenges including the limited number of samples, missing measurements, noisy labels, and heterogeneous data types. Tree-based boosting is well-suited for modeling such data as it is insensitive to data types and missingness. Moreover, Stochastic Gradient TreeBoost is often found in many winning solutions in public data science challenges. Unfortunately, the best performance requires extensive hyperparameter tuning and can be prone to overfitting. We propose PaloBoost, a Stochastic Gradient TreeBoost model that uses novel regularization techniques to guard against overfitting and is robust to hyperparameter settings. PaloBoost uses the out-of-bag samples to perform gradient-aware pruning and estimate adaptive learning rates. Unlike other Stochastic Gradient TreeBoost models that use the out-of-bag samples to estimate test errors, PaloBoost treats the samples as a second batch of training samples to prune the trees and adjust the learning rates. As a result, PaloBoost can dynamically adjust tree depths and learning rates to achieve faster learning at the start and slower learning as the algorithm converges. Experimental results on four datasets demonstrate that PaloBoost is robust to overfitting and is less sensitive to the hyperparameters.

Despite the massive number of enrollments in MOOC (Massive Open Online Course) platforms, dropout rates are very high. This problem can be due to several factors: Social, pedagogical, prior knowledge as well as a demotivation. To deal with this type of problems, we have designed an adaptive cMOOC (Connectivist MOOC) platform for each registered learner’s profile. From the first human-machine interaction, the process adapts the learner's need according to a pre-established model. It is based on the processing of statistical data collected by correspondence analysis and regression algorithms. Each generated learner’s profile will provide an adaptive navigation and pedagogical activities. The intelligent system presented in this work will be able to classify learners according to their preferences and learning styles.

It has been broadly argued that, in the near future, the demand for skilled labor will increase whilst that for routine activities will decrease. In this regard, the need for making greater investments in education to re-skill workers and support continuous learning has been invoked as an essential requirement for preserving people’s employability.Digital technology is deemed increasingly necessary to sustain the educational endeavor, for the possibilities it offers to make more accessible and low-cost educational interventions. It allows for the creation of personalized learning paths and customized digital learning solutions, for courses to be available to a large attendance of learners, and for teaching-learning activities to be offered at significantly reduced cost.In this article, a learning unit structure designed to improve adaptive learning is proposed, and mechanisms for adaptive learning in a smart learning environment are discussed.The implemented teaching-learning solution is also illustrated. This is a preliminary application based on an approach that combines the teacher experience with learning analytics.

An adaptive learning system offers a digital learning environment that adjusts itself to the individual learner and learning material. By refining its internal model of the learner and material over time, such a system continually improves its ability to present appropriate exercises that maximise learning gains. In many cases, there is an initial mismatch between the internal model and the learner’s actual performance on the presented items, causing a “cold start” during which the system is poorly adjusted to the situation. In this study, we implemented several strategies for mitigating this cold start problem in an adaptive fact learning system and experimentally tested their effect on learning performance. The strategies included predicting difficulty for individual learner-fact pairs, individual learners, individual facts, and the set of facts as a whole. We found that cold start mitigation improved learning outcomes, provided that there was sufficient variability in the difficulty of the study material. Informed individualised predictions allowed the system to schedule learners’ study time more effectively, leading to an increase in response accuracy during the learning session as well as improved retention of the studied items afterwards. Our findings show that addressing the cold start problem in adaptive learning systems can have a real impact on learning outcomes. We expect this to be particularly valuable in real-world educational settings with large individual differences between learners and highly diverse materials.

International audience; Network-monitoring data commonly arrives in the form of fast and changing data streams. Continuous and dynamic learning is an effective learning strategy when dealing with such data, where concept drifts constantly occur. We propose different stream-based, adaptive learning approaches to analyze networktraffic streams on the fly. We address two major challenges associated to stream-based machine learning and online network monitoring: (i) how to dynamically learn from and adapt to non-stationary data changing over time, and (ii) how to deal with the limited availability of labeled data to continuously tune a supervised-learning model. We introduce ADAM & RAL, two stream-based machine-learning techniques to tackle these challenges. ADAM relies on adaptive memory strategies to dynamically tune stream-based learning models to changes in the input data distribution. RAL combines reinforcement learning with stream-based active-learning to reduce the amount of labeled data needed for continual learning, dynamically deciding on the most informative samples to learn from. We apply ADAM & RAL to the real-time detection of network attacks in Internet network traffic, and show that it is possible to continuously achieve high detection accuracy even under the occurrence of concept drifts, limiting the amount of labeled data needed for learning.

Scholarly articles are a great source of knowledge. Learning from them like E-learning requires automatic approaches to build concept-maps, learning paths, etc., as these sources are monotonically increasing and are big too. These sources have multi-domain, variety, huge volumes, which are, in fact, Big Data’s characteristics. Thus data from different domains have to be handled together, especially in the E-learning systems. This paper presents a new approach for concept extraction and semantically clustering and classification of these e-learning concepts using fuzzy membership values. Scholarly articles from different domains are taken for our experimental work, and we tested on BBC datasets with 100 documents and 650 documents. Since the number of domains is known, and all concepts are stored, we have done both clustering and classification for testing our fuzzy-based semantic system. We have used logistics regression, Support Vector Machine (SVM) with Linear kernel, Polynomial kernel, Radial Basis Function (RBF) Kernel, Sigmoid kernel to obtain maximum accuracy up to 94% to 96% for all data sets. In clustering, using K-Means, we got precision up to 93%. The system can be used to generate adaptive learning paths, concept map extraction; Big-Data based E-Learning portals.

With an emphasis on learning flexibility, effectiveness, efficiency, engagement, adaptivity and reflectiveness, smart learning embraces a variety of concepts, including but not limited to personalised learning, adaptive learning, intelligent tutoring, open online learning, blended learning, and collaborative learning. As new concepts continue to evolve, the pursuit of smart learning is ongoing, mainly in areas pertaining to the design and implementation frameworks, pedagogical theories and practices, learners’ behaviours and learning pattern, learning and assessment strategies and evaluation of learning performance and perception. This editorial gives an overview of smart learning and provides the context on the latest development of smart learning in which the articles in this special issue are located.

There exist numerous resources online to gain the desired level of knowledge on any topic. However, this complicates the process of selecting the most appropriate resources. Every learner differs in terms of their learning speed, proficiency, and preferred mode of learning. This paper develops an adaptive learning management system to tackle this challenge. It creates a customized course for every student based on their level of knowledge, preferred mode of learning and continuously updates the course based on their learning speed. The material is filtered from a knowledge base that is dynamically updated using web scraping and ranked using feedback from students on the relevance and quality of each material. The model is tested in two phases: the content generation algorithm and the learnability of the system as a whole. The evaluation is done both quantitatively and qualitatively and validated with statistical analysis. Real-time testing of the system shows state-of-the-art performance.

Enhanced richness and size of the data on the web paves the path for increased online services, supporting the sophisticated usage of heterogeneous complex tasks by users. As deep web online services are increasing, several efficient techniques to explore the location of deep web interfaces are also leveraged thereby providing better support in user information exploration on the web concerning queries and the user clicks on web-related search data. This searching process provides an improved experience to users using crawling websites and searches rank with titles and links. The main problem behind search data on the web is organizing user search information based on their preferences into dynamic query formations and data coming from associated web links in an automated way on web services on high-quality sparse data. Vast amounts of data with the Dynamic nature of web interfaces explore or achieve high efficiency and coverage of all deep web interfaces that turn out to be a challenging issue. To handle such issues, we propose and introduce a novel and efficient two-phase deep learning data crawler framework (NTPDCF). The first phase initiates in gathering accurate and highly relevant links using the search engine, and the second phase explores fast and in-site relevant website links using adaptive site ranking. The approach focuses on drilling relevant site data and top-k ranking with different relations based on dynamic features with user preferences in single- and multi-query formation with adaptive weight features. The method promises to visualize improved results with efficient data exploration over the traditional approach concerning real-time defense and e-commerce related to web-based services.

The effective adoption of online learning depends on user satisfaction as distance education approaches suffer from a lack of commitment that may lead to failures and dropouts. The adaptive learning literature argues that an alternative to achieve student satisfaction is to treat them individually, delivering the educational content in a personalized manner. In addition, the sequencing of this content-called Adaptive Curriculum Sequencing (ACS)-is important to avoid cognitive overload and disorientation. The search for an optimal sequence from ever-growing databases is an NP-Hard combinatorial optimization problem. Although some approaches have been proposed, it is challenging to assess their contributions due to the lack of benchmark data available. This paper presents a procedure to create synthetic dataset to evaluate ACS approaches and, as a concept proof, analyzes metaheuristics usually used in ACS approaches: Genetic Algorithm, Particle Swarm Optimization (PSO) and Prey-Predator Algorithm using student's learning goals and their extrinsic and intrinsic information. We also propose an approach based on Differential Evolution (DE). The computational experiments include synthetic datasets with a varied amount of learning materials and real-world datasets for comparison. The results show that DE performed better than the other methods when less than 500 learning materials are used while PSO performed better for larger problems.

Machine learning algorithms are becoming very efficient in intrusion detection systems with their real time response and adaptive learning process. A robust machine learning model can be deployed for anomaly detection by using a comprehensive dataset with multiple attack types. Nowadays datasets contain many attributes. Such high dimensionality of datasets poses a significant challenge to information extraction in terms of time and space complexity. Moreover, having so many attributes may be a hindrance towards creation of a decision boundary due to noise in the dataset. Large scale data with redundant or insignificant features increases the computational time and often decreases goodness of fit which is a critical issue in cybersecurity. In this research, we have proposed and implemented an efficient feature selection algorithm to filter insignificant variables. Our proposed Dynamic Feature Selector (DFS) uses statistical analysis and feature importance tests to reduce model complexity and improve prediction accuracy. To evaluate DFS, we conducted experiments on two datasets used for cybersecurity research namely Network Security Laboratory (NSL-KDD) and University of New South Wales (UNSW-NB15). In the meta-learning stage, four algorithms were compared namely Bidirectional Long Short-Term Memory (Bi-LSTM), Gated Recurrent Units, Random Forest and a proposed Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM) for accuracy estimation. For NSL-KDD, experiments revealed an increment in accuracy from 99.54% to 99.64% while reducing feature size of one-hot encoded features from 123 to 50. In UNSW-NB15 we observed an increase in accuracy from 90.98% to 92.46% while reducing feature size from 196 to 47. The proposed approach is thus able to achieve higher accuracy while significantly lowering number of features required for processing.

Visual diagnosis of radiographs, histology and electrocardiograms lends itself to deliberate practice, facilitated by large online banks of cases. Which cases to supply to which learners in which order is still to be worked out, with there being considerable potential for adapting the learning. Advances in statistical modeling, based on an accumulating learning curve, offer methods for more effectively pairing learners with cases of known calibrations. Using demonstration radiograph and electrocardiogram datasets, the advantages of moving from traditional regression to multilevel methods for modeling growth in ability or performance are demonstrated, with a final step of integrating case-level item-response information based on diagnostic grouping. This produces more precise individual-level estimates that can eventually support learner adaptive case selection. The progressive increase in model sophistication is not simply statistical but rather brings the models into alignment with core learning principles including the importance of taking into account individual differences in baseline skill and learning rate as well as the differential interaction with cases of varying diagnosis and difficulty. The developed approach can thus give researchers and educators a better basis on which to anticipate learners' pathways and individually adapt their future learning.

This paper provides a systematic review of the quality of adaptive learning tools over non-adaptive learning tools. A search using the relevant keywords in Google Scholar yielded 66 usable papers. These were categorised based on the works and the countries where most of the research has taken place. Perceived problems were identified and a framework / model as a suitable solution was developed - both adaptive and non-adaptive. Adaptive learning is an interactive experience which leverages technology and provides a responsive teacher to the student. The advantages over non-adaptive methods are numerous and mentioned here. A review of these showed their superiority also in terms of quality over the existing methods. Learning styles are the prominent points of discussion and form the basis for developing frameworks or models. This research has implications on the quality of learning capability of the students and also on the quality of delivery of teaching. Infrastructural limitations in schools are possible but can be overcome through smartphones / other technologies. For future research, it would be worthwhile to compare the adaptive learning tools with non-adaptive learning tools using a quantitative methodology.

Objective Providing behavioral health interventions via smartphones allows these interventions to be adapted to the changing behavior, preferences, and needs of individuals. This can be achieved through reinforcement learning (RL), a sub-area of machine learning. However, many challenges could affect the effectiveness of these algorithms in the real world. We provide guidelines for decision-making. Materials and Methods Using thematic analysis, we describe challenges, considerations, and solutions for algorithm design decisions in a collaboration between health services researchers, clinicians, and data scientists. We use the design process of an RL algorithm for a mobile health study “DIAMANTE” for increasing physical activity in underserved patients with diabetes and depression. Over the 1.5-year project, we kept track of the research process using collaborative cloud Google Documents, Whatsapp messenger, and video teleconferencing. We discussed, categorized, and coded critical challenges. We grouped challenges to create thematic topic process domains. Results Nine challenges emerged, which we divided into 3 major themes: 1. Choosing the model for decision-making, including appropriate contextual and reward variables; 2. Data handling/collection, such as how to deal with missing or incorrect data in real-time; 3. Weighing the algorithm performance vs effectiveness/implementation in real-world settings. Conclusion The creation of effective behavioral health interventions does not depend only on final algorithm performance. Many decisions in the real world are necessary to formulate the design of problem parameters to which an algorithm is applied. Researchers must document and evaulate these considerations and decisions before and during the intervention period, to increase transparency, accountability, and reproducibility. Trial Registration clinicaltrials.gov, NCT03490253.

Objective: Several features of the surface electromyography (sEMG) signal are related to muscle activity and fatigue. However, the time-evolution of these features are non-stationary and vary between subjects. The aim of this study is to investigate the use of adaptive algorithms to forecast sEMG feature of the trunk muscles. Methods: Shallow models and a deep convolutional neural network (CNN) were used to simultaneously learn and forecast 5 common sEMG features in real-time to provide tailored predictions. This was investigated for: up to a 25 second horizon; for 14 different muscles in the trunk; across 13 healthy subjects; while they were performing various exercises. Results: The CNN was able to forecast 25 seconds ahead of time, with 6.88% mean absolute percentage error and 3.72% standard deviation of absolute percentage error, across all the features. Moreover, the CNN outperforms the best shallow model in terms of a figure of merit combining accuracy and precision by at least 30% for all the 5 features. Conclusion: Even though the sEMG features are non-stationary and vary between subjects, adaptive learning and forecasting, especially using CNNs, can provide accurate and precise forecasts across a range of physical activities. Significance: The proposed models provide the groundwork for a wearable device which can forecast muscle fatigue in the trunk, so as to potentially prevent low back pain. Additionally, the explicit real-time forecasting of sEMG features provides a general model which can be applied to many applications of muscle activity monitoring, which helps practitioners and physiotherapists improve therapy.

POCI-01-0145-FEDER-030433 LISBOA-01-0145-FEDER-0307095 This work presents an innovative methodology to predict the future trajectories of vehicles when its current and previous locations are known. We propose an algorithm to adapt the vehicles trajectories' data based on consecutive GPS locations and to construct a statistical inference module that can be used online for mobility prediction. The inference module is based on a hidden Markov model (HMM), where each trajectory is modeled as a subset of consecutive locations. The prediction stage uses the statistical information inferred so far and is based on the Viterbi algorithm, which identifies the subset of consecutive locations (hidden information) with the maximum likelihood when a prior subset of locations are known (observations). By analyzing the disadvantages of using the Viterbi algorithm (TDVIT) when the number of hidden states increases, we propose an enhanced algorithm (OPTVIT), which decreases the prediction computation time. Offline analysis of vehicle mobility is conducted through the evaluation of a dataset containing real traces of 442 taxis running in the city of Porto, Portugal, during a full year. Experimental results obtained with the dataset show that the prediction process is improved when more information about prior vehicle mobility is available. Moreover, the computation time of the prediction process is significantly improved when OPTVIT is adopted and approximately 90% of prediction performance can be achieved, showing the effectiveness of the proposed method for vehicle trajectory prediction. publishersversion published

Autonomous driving (AD) provides a reliable solution for safe driving by replacing human drivers responsible for the majority of accidents. The emergence of Machine Learning, specifically Deep Reinforcement Learning (DRL), and its ability to solve complex games proved its potential to address AD challenges. However, model-free methods still suffer from safety-related issues that can be resolved using safe-DRL approaches. The addition of model-based safety filters to the learning-based algorithms provides safety bounds on their performance and constraint satisfaction. In this paper, we investigate the addition of a safety filter based on Model Predictive Control and show an increase in mean testing episode reward by 110% from −75 mean episode reward during testing for 50 episodes for Deep Deterministic Policy Gradient ( $DDPG$ ) to 7.758. We study the impacts of safety filters (7.758 mean reward), heuristic rules, bounded additive noises (0.49% performance increase comparing to noise-free case), and exploration (3.425 mean reward) on the learning algorithm. We compare the effects of filters in the context of simulated exploration and bounded exploration and prove that bounded exploration results in 9.86% increase in mean reward and 12.95% decrease in std comparing to the other method. Additionally, inspired by Deep Internal Learning and biological mechanisms like brain plasticity, we investigate the idea of using each sample for training only once instead of utilizing stochastic batches which increases the mean testing accumulated reward by 1.87% and leads to the best performance (7.942 mean reward and 0.048 std). Finally, the results demonstrate better automotive results for our proposed method than DDPG. Our proposed method, DDPG with safety filter in bounded exploration and adaptive learning under noisy input conditions, has a success rate of 100% under different traffic densities for the simulation environment used in this paper and our assumptions. The proposed method’s automotive results are shown for a braking scenario to avoid collision with other road users.

Amid the recent disease outbreaks that have been spreading across the world, the education systems in every country have witnessed a dramatic transformation. In particular, the situation has promoted online learning on an unprecedented scale, with classes being held virtually on digital platforms. This vital transformation is a major challenge in developing countries where information infrastructure for remote learning is lacking. Inspired by these observations, this study first investigated the obstacles in the way of the deployment of smart education systems (SESs) in developing countries from a technical perspective. Consequently, a detachable SES framework, named vSmartEdu, is proposed. The framework is based on a hybrid online/offline web-service model, which adopts a service-based architecture (SBA) design concept to develop smart classrooms. In particular, the online mode is activated for a web-based classroom if an Internet service is available. In contrast, the offline-version of the system is available in offline mode in packaged form, and is utilized when the Internet is not available in a local classroom. Finally, a prototype was deployed to collect feedback from learners and educators at various educational levels. The trial implementation and survey results concretely validate the feasibility and advantages of the proposed solution.

Imaging quality judgement provides useful meaning to benefit intelligent visual power systems. However, accurate identification of possible failures of the monitoring equipment remains challenging. This paper proposes a novel deep architecture to improve the assessment of abnormalities by considering a novel mutual-guided convolutional attention network (MGCAN) and a multi-region scheme. More specifically, the MGCAN exploits various adequate, low-level information to generate spatial and structural flows. Accordingly, it further extends attention mechanisms to model both inter-flow correlations and inter-channel relationships in the network design. The multi-region scheme features a spatial pyramid random-crop strategy and a region-fusion strategy to handle locally non-uniform characteristics among categories. In this way, the whole architecture provides an end-to-end and adaptive learning procedure relevant to quality perception that focuses on learning important features and mining discriminative regions. Experimental results demonstrate its superiority to prior methods for judging abnormalities. The proposed method can be easily extended to an entire surveillance system.

In the existing social recommendation algorithms, the performance of the algorithms is bothered by the problem of noise and sparsity in social information. In the past few years, researchers have usually focused on solving noise problem or sparse problem, but ignored a key problem, that is if only solving the noise problem, the sparse problem will become more serious, similarly, if only solving the sparse problem, the noise problem will become more serious. Recently, a few studies attempt to solve both noise and sparse problems, usually using explicit trust features to filter trust information and generate implicit trust friends, but this will depend heavily on the quality of explicit trust features. Therefore, in this paper, we propose an adaptive algorithm based on graph convolution network, The main idea is to learn users’ implicit trust behavior adaptively. Briefly, we use Jaccard similarity coefficient to filter out unnecessary trust information, the reliable trust information will be the users’ trust attributes, then put into the graph convolution network, because of the good feature extraction ability of graph convolution network, we can use adaptive network to learn implicit trust features based on users’ features, the trust features can infer the trust attributes for all users. In order to achieve the best effect of learning, we use iterative repetition of the above process, to mining the implicit influence of user trust behavior, so as to alleviate noise and sparsity problems. Experimental result shows that, on three open datasets: LastFM, Douban and Gowalla, the recommendation performance is improved, which proves the effectiveness of the algorithm.

Cyber-manufacturing relies on smart digital-twins of manufacturing processes that can quickly act for making a wise decision. However, the cognitive computing part of the digital-twin becomes time-intensive beyond the requirement of a smart system when it uses simulation tools that solve governing constitutive equations in the form of partial differential equations (PDE). On the other hand, many artificial intelligence (AI) and machine learning (ML) solutions rely on a large data set that does not exist in many manufacturing systems. We build a hybrid digital-twin that takes advantage of an ML-based digital-twin for quick response while gaining fidelity through adaptive learning with a PDE-based digital-twin. We use our hybrid digital-twin for active exploration of various overlay metal deposition patterns in real-time. This tool enables engineers to explore and compare many patterns they need to assess metal deposition scenarios with no delay for computational time.

Distributed systems have problems with downtime, data loss during malfunctions, scalability and efficient use of computing resources. At the same time in the learning and training process, the use of a distributed system has the advantage of data processing: storage of information about students, construction of training courses, verification of passed material, etc. The problems of scaling and efficient use of resources in distributed learning management systems are investigated in this research. Cloud platforms for hosting the system, such as Amazon Web Services, Microsoft Azure, Google Cloud Platform and DigitalOcean are reviewed. Problems and features of a scalability in cloud computing are discussed. Methods, scaling and load balancing algorithms for the efficient use of computing resources are proposed. According to the list of advantages, the DigitalOcean platform was selected for the investigation. DigitalOcean provides cloud servers that can be used for quick creation of the new virtual machines for the projects. These servers allow to fully control the web hosting environment at the same time that the user pays only for the resources used. The main goal of DigitalOcean is to use a solid-state drive (SSD) to create a user-friendly platform that will allow clients to migrate projects to and from the cloud, increasing productivity with high speed and efficiency. As a result of analyzing information on existing technologies, approaches and methods for using cloud platforms in distributed systems, they have been applied to develop a solution to reduce downtime for a distributed adaptive Learning Management System (LMS). It is concluded that the use of cloud platforms for the construction of distributed LMS a practice that allows to use only the required amount of computing capacity. It is proven, that the implementation of the proposed solution into the work of adaptive LMS will improve its efficiency by reducing the time of the content delivering. Розподілені системи мають проблеми з простоєм, втратою даних під час несправностей, масштабованістю та ефективним використанням обчислювальних ресурсів. Водночас у процесі навчання та тренінгу використання розподіленої системи має перевагу обробки даних: зберігання інформації про студентів, побудова навчальних курсів, перевірка зданого матеріалу тощо. У даній роботі досліджено проблеми масштабування та ефективного використання ресурсів в розподілених системах управління навчанням. Розглянуто хмарні платформи для розміщення системи, такі як Amazon Web Services, Microsoft Azure, Google Cloud Platform та DigitalOcean. Розглянуто проблеми та особливості масштабованості хмарних обчислень. Запропоновано методи, алгоритми масштабування та балансування навантаження для ефективного використання обчислювальних ресурсів. Відповідно до переліку переваг, для дослідження було обрано платформу DigitalOcean. DigitalOcean надає хмарні сервери, які можна використовувати для швидкого створення нових віртуальних машин для проектів. Ці сервери дозволяють повністю контролювати середовище веб-хостингу, при цьому користувач платить лише за використані ресурси. Основною метою DigitalOcean є використання твердотільного накопичувача (SSD) для створення зручної платформи, яка дозволить клієнтам перемішувати проекти в хмару та з неї, підвищуючи продуктивність з високою швидкістю та ефективністю. В результаті аналізу інформації про існуючі технології, підходи та методи використання хмарних платформ у розподілених системах було розроблено рішення щодо скорочення простоїв розподіленої адаптивної системи управління навчанням (LMS). Зроблено висновок про те, що використання хмарних платформ для побудови розподілених LMS – це практика, яка дозволяє використовувати лише необхідну кількість обчислювальних потужностей. Доведено, що впровадження запропонованого рішення в роботу адаптивної LMS підвищить її ефективність за рахунок скорочення часу доставки контенту.

Predictive models operating on the evolving data streams are dynamic. The performance of a model will deteriorate eventually when it suffers the effect of concept drift. The learning algorithms require the proper adaptive strategies to cope with the concept drifting data streams. In this paper, we propose a new concept drift detection approach that analyzes the consistency of prequential error rate using Hoeffding’s inequality to detect the concept drifts in data streams. The experimental results show that our proposed method outperforms in comparison with other state-of-the-art detectors in terms of true drift detection, false alarm and delay of detection.

Adaptive Learning (AL), a new web-based online learning environment, requires self-regulated learners who act autonomously. However, to date, there appears to be no existing model to conceptualize different aspects of SRL skills in Adaptive Learning Environments (ALE). The purpose of this study was to design and empirically evaluate a theoretical model of Self-Regulated Learning (SRL) in ALE's and the related questionnaire as a measurement tool. The proposed theoretical model, namely, “Adaptive Self-Regulated Learning (ASR)”, was specified to incorporate the SRL skills into ALE's. Based on this model, the Adaptive Self-regulated Learning Questionnaire (ASRQ) was developed. The reliability and validity of the ASRQ were evaluated via the review of a content expert panel, the Cronbach's alpha coefficients, and confirmatory factor analysis. Overall, the results supported the theoretical framework and the new ASRQ in an ALE. In this article, the theoretical and practical implications of the findings were discussed.

To achieve adaptive learning, a dynamic assessment system equipped with a cognitive diagnosis was developed for this study, which adopts a three-stage model of diagnosis-intervention-assessment. To...

Text Classification is a fundamental and crucial issue in many Natural Language Processing (NLP) tasks. An effective and efficient representation model is the key to text classification. However, most existing representation models either learn insufficient structural information or just rely on pre-defined structures, resulting in degradation of performance and generalization capability. We propose a novel Sandwich Neural Network (SNN), which is able to learn local semantic and global structural representations automatically without parsers. To combine semantic and structural representations sensibly, we propose four fusion methods incorporated with SNN: Static Fusion, Adaptive Learning, Self-Attention, and Knowledge Attention methods. Static Fusion weights semantic and structural representations equally, Adaptive Learning learns the weights at corpus level, and Self-Attention learns the weights at instance level. More importantly, within Knowledge Attention fusion method, external semantic and structural knowledge are incorporated into SNN to improve the attention procedure and further boost the performance of SNN. Evaluated with four mainstream datasets: Text REtrieval Conference (TREC), SUBJectivity (SUBJ), Movie Reviews (MR) and Stanford Sentiment Treebank with Five Labels (SST-5), the proposed Knowledge Attention Sandwich Neural Network(KA-SNN) model achieves very competitive performance. Specifically, the accuracies achieve 96.2% (TREC), 94.1% (SUBJ), 82.3% (MR) and 51.5% (SST-5). Moreover, the proposed Knowledge Attention reduces the structural complexity of attention module by 77.66-81.43% with a computing time reduction of 21.47-34.05%, compared with Self-Attention fusion method.

Smart Learning Environment (SLE) is the hourly need to satisfy the urge generated by the new knowledge society. This includes a focus on learning methods to yield a required blend of learning environment that supports experience of effective and engaged learning. The essence of ‘SLE’ is to create a condition that can enhance the learners’ personalized and adaptive learning experience. For the Architecture and Design students, Studio- a physical space that allows students to accomplish their design learning in an innovative and creative manner, plays an integral role in their learning. The main objective of the paper is to redefine the concept of ‘Smart Learning Environment’ in the context of Design education and suggest a model for the ‘Smart Design Studio’ based on the perception of learners. The methodology includes surveys and group discussion at the University of Bahrain (UoB), to receive opinions on smart teaching-learning spaces and methods. A short experimental study was done to measure the impact on the design learning as a result of modifications done in the Design Studio. The experiment was centered on the theory of ‘Identity Development’ and ‘Personalization of spaces’. Finally, the paper concludes a new vision on SLE as a learner-centered-environment that allows multiple ways of learning to support all-round development of the learners. As a recommendation, authors suggest a model for the Design Studio to achieve effective, engaged and efficient learning. The key elements of the model are the disseminator and recipient of knowledge, learning methods, modes, tools and resources.

Objectives: We survey recent developments in medical Information Extraction (IE) as reported in the literature from the past three years. Our focus is on the fundamental methodological paradigm shift from standard Machine Learning (ML) techniques to Deep Neural Networks (DNNs). We describe applications of this new paradigm concentrating on two basic IE tasks, named entity recognition and relation extraction, for two selected semantic classes—diseases and drugs (or medications)—and relations between them.Methods: For the time period from 2017 to early 2020, we searched for relevant publications from three major scientific communities: medicine and medical informatics, natural language processing, as well as neural networks and artificial intelligence.Results: In the past decade, the field of Natural Language Processing (NLP) has undergone a profound methodological shift from symbolic to distributed representations based on the paradigm of Deep Learning (DL). Meanwhile, this trend is, although with some delay, also reflected in the medical NLP community. In the reporting period, overwhelming experimental evidence has been gathered, as illustrated in this survey for medical IE, that DL-based approaches outperform non-DL ones by often large margins. Still, small-sized and access-limited corpora create intrinsic problems for data-greedy DL as do special linguistic phenomena of medical sublanguages that have to be overcome by adaptive learning strategies.Conclusions: The paradigm shift from (feature-engineered) ML to DNNs changes the fundamental methodological rules of the game for medical NLP. This change is by no means restricted to medical IE but should also deeply influence other areas of medical informatics, either NLP- or non-NLP-based.

The minority oversampling techniques have substantiated their appropriateness and utility in the domain of class-imbalance learning. However, this does not affirm the true class of the synthetic minority points. In this work, Adaptive Learning of Minority Class prior to Minority Oversampling (ALMCMO), we work towards bridging this gap by estimating the minority set before oversampling the synthetic points. We estimate a varying and adaptive volume of minority space around the minority points. We aim to guarantee the class-memberships of the synthetic minority points by sampling them from the estimated minority spaces. In our empirical study, we have used six comparing methods, 23 datasets and two classifiers. The results indicate the certain superiority of the proposed method over the six competing schemes.

Knowledge tracing (KT) is essential for adaptive learning to obtain learners’ current states of knowledge for the purpose of providing adaptive service. Generally, the knowledge construction procedure is constantly evolving because students dynamically learn and forget over time. Unfortunately, to the best of our knowledge most existing approaches consider only a fragment of the information that relates to learning or forgetting, and the problem of making use of rich information during learners’ learning interactions to achieve more precise prediction of learner performance in KT remains under-explored. Moreover, existing work either neglects the problem difficulty or assumes that it is constant, and this is unrealistic in the actual learning process as problem difficulty affects performance undoubtedly and also varies overtime in terms of the cognitive challenge it presents to individual learners. To this end, we herein propose a novel model, KTM-DLF (Knowledge Tracing Machine by modeling cognitive item Difficulty and Learning and Forgetting), to trace the evolution of each learner’s knowledge acquisition during exercise activities by modeling his or her dynamic knowledge construction procedure and cognitive item difficulty. Specifically, we first specify the concept of cognitive item difficulty and propose a method to model the cognitive item difficulty adaptively based on learners’ learning histories. Then, based on two classical theories (the learning curve theory and the Ebbinghaus forgetting curve theory), we propose a method for modeling learners’ learning and forgetting over time. Finally, the KTM-DLF model is proposed to incorporate learners’ abilities, the cognitive item difficulty, and the two dynamic procedures (learning and forgetting) together. We then use the factorization machine framework to embed features in high dimensions and model pairwise interactions to increase the model’s accuracy. Extensive experiments have been conducted on three public real-world datasets, and the results confirm that our proposed model outperforms the other state-of-the-art educational data mining models.

This article proposes a computer-based approach to effectively enhance second language learners’ willingness to communicate in the target language. To do so, we implemented a conversational agent embedding a dialogue management model based on two conversational strategies (i.e., communication strategies and affective backchannels), serving as scaffolds for enhancing learners’ willingness to communicate in the target language. Here, we report on differences observed among second language learners’ preferences for both conversational strategies according to their initial level of willingness to communicate and on variations of their willingness with respect to such differences. Although we found that most students generally preferred a combination of both strategies, learners’ preferences and the effects of the support provided by these strategies varied according to their level of willingness to communicate. Learners with lower willingness to communicate tended to prefer affective backchannels, whereas those with higher willingness to communicate seemed to favor communication strategies. These results were consistent with post-test results, which showed that learners’ expected willingness to communicate tended to be higher after interacting with systems embedding their preferred strategies. In sum, these results are preliminary evidence of the meaningfulness of accounting for such learners’ preferences in adaptively using and fading the strategies employed by conversational agents to motivate second language learners to communicate in the target language.

Online education and e-learning have vigorously sprung up and produced massive educational data in a streaming way. It is very challenging to acquire the appropriate learning resources and the suitable learning partners from the streaming-updated educational big data. This article aims to provide sustainable smart educational services including precise personalized recommendation and adaptive clustering under different contexts by correlatively analyzing the global educational data from multiple dimensions via incremental tensor decomposition. First, a tensor-based recommendation and service framework for the streaming educational big data is developed. Then three local tensors concerning learners, resources, and learning records are constructed and further fused to an integrated global learner-resource tensor. Afterward, we present an incremental tensor-based correlative analysis and personalized recommendation (ITCA-PR) algorithm to recommend appropriate resources under various contexts. Besides, we also propose an incremental tensor-based adaptive clustering and community recommendation (ITAC-CR) algorithm to recommend suitable learning partners under various contexts and accordingly construct adaptive learning communities. Extensive experimental results demonstrate that the ITCA-PR algorithm outperforms state-of-the-art improved collaborative filtering algorithms by F-score measurement and the ITAC-CR algorithm has a better clustering performance by DVI measurement. These precise educational services will promote the development of sustainable smart education.