Dynamic Learning Research Articles

FairDPFL-SCS: Fair Dynamic Personalized Federated Learning with strategic client selection for improved accuracy and fairness

Personalized federated learning (PFL) addresses the significant challenge of non-independent and identically distributed (non-IID) data across clients in federated learning (FL). Our proposed framework, “FairDPFL-SCS: Fair Dynamic Personalized Federated Learning with Strategic Client Selection”, marks a notable advancement in this domain. By integrating dynamic learning rate adjustments and a strategic client selection mechanism, our approach effectively mitigates the challenges posed by non-IID data while enhancing model personalization, fairness, and efficiency. We evaluated FairDPFL-SCS using standard datasets, including MNIST, FashionMNIST, and SVHN, employing architectures like VGG and CNN. Our model achieved impressive results, attaining 99.04% accuracy on MNIST, 89.19% on FashionMNIST, and 90.9% on SVHN. These results represent a substantial improvement over existing methods, including a highest increase of 16.74% in accuracy on SVHN when compared to the best-performing benchmark methods. In particular, our method also demonstrated lower fairness variance, presenting the importance of fairness in model personalization, a frequently overlooked aspect in FL research. Through extensive experiments, we validate the superior performance of FairDPFL-SCS compared to benchmark PFL approaches, highlighting significant improvements over state-of-the-art methods. This work represents a promising step forward in the field of federated learning, offering a comprehensive solution to the challenges presented by non-IID data while prioritizing fairness and efficiency in model personalization.

Influence of App Usability and Learning Environment on Student Behavior in Mobile-Based Marketing Education in Malaysian Higher Learning Institutions

The increasing integration of mobile-based platforms in higher education, particularly within marketing education, has transformed learning dynamics by enabling students to engage through accessible digital tools. However, while internet and smartphone usage continue to grow in Malaysia, student engagement levels with mobile learning apps remain inconsistent. This study investigates how app usability (ease of use) and the learning environment influence student engagement behaviors within mobile-based marketing education in Malaysian higher education institutions. Using a quantitative approach, data were collected through a survey of 115 marketing students from public and private universities in Malaysia who have experience learning through mobile apps. The analysis, conducted using SPSS, indicates that both app usability and a supportive learning environment significantly contribute to positive student engagement behaviors, such as increased participation. The findings highlight that an intuitive, easy-to-use mobile platform within an encouraging educational environment fosters higher engagement in digital learning settings. This research provides valuable insights for educators and developers of mobile learning tools, underlining the need to prioritize usability and environmental support to optimize student engagement. The study’s implications can inform strategies for enhancing student interactions with digital marketing content, offering practical guidance for stakeholders aiming to elevate mobile-based educational experiences in Malaysia’s higher education sector.

Metaverse Applications in Education 4.0: A Decade of Systematic Literature Review

The advent of Education 4.0, which aligns with the fourth industrial revolution, has been significantly influenced by advancements in digital technologies. Central to this evolution is the metaverse, a virtual shared space that merges augmented reality, virtual reality, and physical reality. This paper delves into the metaverse's applications within Education 4.0, highlighting its potential to revolutionize learning experiences, enhance collaboration, and improve access to quality education. By reviewing current literature and case studies, we identify the primary benefits, such as increased engagement, personalized learning, and broader accessibility. Additionally, we address the challenges associated with metaverse integration, including technical limitations, privacy concerns, and the need for new pedagogical approaches. Through a mixed-methods research approach, combining qualitative and quantitative data along with expert interviews, this paper provides a comprehensive overview of the metaverse's role in future education. The findings suggest that while significant hurdles remain, the metaverse offers a promising avenue for transforming educational practices to meet the demands of the digital age, ultimately fostering more dynamic, inclusive, and effective learning environments. Future research should focus on evaluating long-term impacts and developing standards for metaverse applications in education.

Using computational fluid dynamics and deep learning for leak detection and localization in a smart water management system

Water leakages in distribution systems offer significant challenges, resulting in water loss, infrastructure damage, and environmental hazards. As urban populations grow and water resources become scarcer, water management systems’ efficiency becomes increasingly important. Traditional leak detection methods are inadequate and labour-intensive, making them unsuitable for large-scale infrastructure. In addition, limitations in the data representing pipes of varying diameters and materials under various leak scenarios leads to complexity in automatic water leak detection and localization. In response to this challenge, this research effort combines Computational Fluid Dynamics (CFD) and Deep Learning (DL) to enhance leak detection and localization. The study simulates water flow in pipes of varying diameters and materials under various leak scenarios using CFD. Consequentially deep learning models like the Bi-Layered ReLU Encoder + Softmax and Tri-Layered ReLU Encoder + Softmax are trained and tested. The results indicate improvements in detection accuracy by 2.4% and localization by 5.6%, highlighting the potential of this hybrid approach to enhance the reliability and efficiency of water management systems, ultimately promoting water conservation and reducing infrastructure damage.

Developing the KATAKU Mobile Application to Enhance Vocabulary Acquisition for Deaf Students

This study presents the development of the KATAKU (Kata untuk Tunarungu Interaktif dan Unik) application as an educational tool designed to support vocabulary learning for deaf students. Deaf students, who have the same educational rights as their hearing peers, often attend Special Schools (SLB). The integration of visual aids and technology in learning has proven effective in increasing their engagement and interest. This study employed a mixed-methods approach, including vocabulary assessments for deaf students, classroom observations, and teacher interviews for a needs analysis to identify the specific requirements of schools and students. Results from the vocabulary assessments revealed that deaf students face significant challenges, particularly with academic terms and word categorization. Classroom observations at three SLB schools in Jombang highlighted engaging but challenging learning dynamics, reinforcing the need for accessible and efficient teaching tools for educators. Teacher interviews indicated that using the paper dictionary provided by the Ministry of Education can be time-consuming and lacks efficiency. This study concludes that deaf students in Jombang face substantial obstacles in vocabulary comprehension, especially with academic vocabulary. The findings underscore the need for targeted vocabulary interventions using modern technological tools to better prepare students for advanced educational demands.

Dynamic Deep Learning to Predict Mechanical Properties of High-Strength Low-Alloy Steels

Dynamic Deep Learning to Predict Mechanical Properties of High-Strength Low-Alloy Steels

ENSINO DE ÂNGULOS E POLÍGONOS COM USO DO SOFTWARE GEOGEBRA

The objective of this work is to describe a didactic-pedagogical sequence developed to teach angles and polygons in an 8th grade class of Elementary School of a state public school in Palmeira dos Índios, Alagoas, Brazil. The sequence is part of the Extension Project "Mathematics and Active Methods" and addresses some common difficulties in teaching geometry, such as the abstraction of formal concepts and the lack of understanding of students of the connection of these concepts with real situations. Studies indicate that the use of digital technologies facilitates visualization and interaction with mathematical concepts, promoting more dynamic and meaningful learning. In view of this, the sequence, applied in four classes, involved activities contextualized to the sport, using the GeoGebra software. The results show that, although some students still face difficulties, the interactive and contextualized approach improved engagement and understanding of formal mathematical concepts, especially geometric ones.

Gas Pipeline Leak Detection by Integrating Dynamic Modeling and Machine Learning Under the Transient State

This study focused on developing machine learning models to detect leak size and location in transient state conditions. The model was designed for an onshore methane–hydrogen blending gas pipeline in Canada. Base case simulations revealed significant effects on mass flow and pressure due to leaks, with the system taking approximately 6 h to reach a steady state from transient conditions. This made it essential to analyze the flow characteristics during the transient state. Trend data from the pipeline’s inlet and outlet were examined, considering the leak size and location. To better represent the data over time, a method was used to create two-dimensional images, which were then fed into a CNN (convolutional neural network) for training. The model’s accuracy was assessed using classification accuracy and a confusion matrix. By refining the data acquisition process and implementing targeted screening procedures, the model’s classification accuracy increased to over 80%. In conclusion, this study demonstrates that machine learning can enable rapid and accurate leak detection in transient state conditions. The findings are expected to complement existing leak detection methods and support operators in making faster, more informed decisions.

Utilizing machine learning and molecular dynamics for enhanced drug delivery in nanoparticle systems

Materials data science and machine learning (ML) are pivotal in advancing cancer treatment strategies beyond traditional methods like chemotherapy. Nanotherapeutics, which merge nanotechnology with targeted drug delivery, exemplify this advancement by offering improved precision and reduced side effects in cancer therapy. The development of these nanotherapeutic agents depends critically on understanding nanoparticle (NP) properties and their biological interactions, often analyzed through molecular dynamics (MD) simulations. This study enhances these analyses by integrating ML with MD simulations, significantly improving both prediction accuracy and computational efficiency. We introduce a comprehensive three-stage methodology for predicting the solvent-accessible surface area (SASA) of NPs, which is crucial for their therapeutic efficacy. The process involves training an ML model to forecast the many-body tensor representation (MBTR) for future time steps, applying data augmentation to increase dataset realism, and refining the SASA predictor with both augmented and original data. Results demonstrate that our methodology can predict SASA values 299 time steps ahead with a 40-fold speed improvement and a 25% accuracy increase over existing methods. Importantly, it provides a 300-fold increase in computational speed compared to traditional simulation techniques, offering substantial cost and time savings for nanotherapeutic research and development.

Machine Learning-Based Modeling for Structural Engineering: A Comprehensive Survey and Applications Overview

Modeling and simulation have been extensively used to solve a wide range of problems in structural engineering. However, many simulations require significant computational resources, resulting in exponentially increasing computational time as the spatial and temporal scales of the models increase. This is particularly relevant as the demand for higher fidelity models and simulations increases. Recently, the rapid developments in artificial intelligence technologies, coupled with the wide availability of computational resources and data, have driven the extensive adoption of machine learning techniques to improve the computational accuracy and precision of simulations, which enhances their practicality and potential. In this paper, we present a comprehensive survey of the methodologies and techniques used in this context to solve computationally demanding problems, such as structural system identification, structural design, and prediction applications. Specialized deep neural network algorithms, such as the enhanced probabilistic neural network, have been the subject of numerous articles. However, other machine learning algorithms, including neural dynamic classification and dynamic ensemble learning, have shown significant potential for major advancements in specific applications of structural engineering. Our objective in this paper is to provide a state-of-the-art review of machine learning-based modeling in structural engineering, along with its applications in the following areas: (i) computational mechanics, (ii) structural health monitoring, (iii) structural design and manufacturing, (iv) stress analysis, (v) failure analysis, (vi) material modeling and design, and (vii) optimization problems. We aim to offer a comprehensive overview and provide perspectives on these powerful techniques, which have the potential to become alternatives to conventional modeling methods.

Skill matters: Dynamic skill learning for multi-agent cooperative reinforcement learning

With the popularization of intelligence, the necessity of cooperation between intelligent machines makes the research of collaborative multi-agent reinforcement learning (MARL) more extensive. Existing approaches typically address this challenge through task decomposition of the environment or role classification of agents. However, these studies may rely on the sharing of parameters between agents, resulting in the homogeneity of agent behavior, which is not effective for complex tasks. Or training that relies on external rewards is difficult to adapt to scenarios with sparse rewards. Based on the above challenges, in this paper we propose a novel dynamic skill learning (DSL) framework for agents to learn more diverse abilities motivated by internal rewards. Specifically, the DSL has two components: (i) Dynamic skill discovery, which encourages the production of meaningful skills by exploring the environment in an unsupervised manner, using the inner product between a skill vector and a trajectory representation to generate intrinsic rewards. Meanwhile, the Lipschitz constraint of the state representation function is used to ensure the proper trajectory of the learned skills. (ii) Dynamic skill assignment, which utilizes a policy controller to assign skills to each agent based on its different trajectory latent variables. In addition, in order to avoid training instability caused by frequent changes in skill selection, we introduce a regularization term to limit skill switching between adjacent time steps. We thoroughly tested the DSL approach on two challenging benchmarks, StarCraft II and Google Research Football. Experimental results show that compared with strong benchmarks such as QMIX and RODE, DSL effectively improves performance and is more adaptable to difficult collaborative scenarios.

Dynamic learning experiences for students

Dynamic learning experiences for students

Distinct mechanisms for online and offline motor skill learning across human development.

The human central nervous system (CNS) undergoes tremendous changes from childhood to adulthood and this may affect how individuals at different stages of development learn new skills. Here, we studied motor skill learning in children, adolescents, and young adults to test the prediction that differences in the maturation of different learning mechanisms lead to distinct temporal patterns of motor learning during practice and overnight. We found that overall learning did not differ between children, adolescents, and young adults. However, we demonstrate that adult-like skill learning is characterized by rapid and large improvements in motor performance during practice (i.e., online) that are susceptible to forgetting and decay over time (i.e., offline). On the other hand, child-like learning exhibits slower and less pronounced improvements in performance during practice, but these improvements are robust against forgetting and lead to gains in performance overnight without further practice. The different temporal dynamics of motor skill learning suggest an engagement of distinct learning mechanisms in the human CNS during development. In conclusion, adult-like skill learning mechanisms favor online improvements in motor performance whereas child-like learning mechanisms favors offline behavioral gains. RESEARCH HIGHLIGHTS: Many essential motor skills, like walking, talking, and writing, are acquired during childhood, and it is colloquially thought that children learn better than adults. We investigated dynamics of motor skill learning in children, adolescents, and young adults. Adults displayed substantial improvements during practice that was susceptible to forgetting over time. Children displayed smaller improvements during practice that were resilient against forgetting. The distinct age-related characteristics of these processes of acquisition and consolidation suggest that skill learning relies on different mechanisms in the immature and mature central nervous system.

Tensor Learning Meets Dynamic Anchor Learning: From Complete to Incomplete Multiview Clustering.

Multiview clustering (MVC), which can dexterously uncover the underlying intrinsic clustering structures of the data, has been particularly attractive in recent years. However, previous methods are designed for either complete or incomplete multiview only, without a unified framework that handles both tasks simultaneously. To address this issue, we propose a unified framework to efficiently tackle both tasks in approximately linear complexity, which integrates tensor learning to explore the inter-view low-rankness and dynamic anchor learning to explore the intra-view low-rankness for scalable clustering (TDASC). Specifically, TDASC efficiently learns smaller view-specific graphs by anchor learning, which not only explores the diversity embedded in multiview data, but also yields approximately linear complexity. Meanwhile, unlike most current approaches that only focus on pair-wise relationships, the proposed TDASC incorporates multiple graphs into an inter-view low-rank tensor, which elegantly models the high-order correlations across views and further guides the anchor learning. Extensive experiments on both complete and incomplete multiview datasets clearly demonstrate the effectiveness and efficiency of TDASC compared with several state-of-the-art techniques.

Social learning and expectational stability

Stability features of social learning (SL) dynamics are examined. We show SL can be formulated as a stochastic recursive algorithm, making it possible to analyze asymptotics using the familiar differential-equation approach. For a simple univariate model, this approach reduces to the E-stability principle, though in prominent instability cases divergence is exceedingly slow compared to adaptive learning (AL). We locate differing fitness criteria as the source of the slower evolution rates of SL compared to AL. Modified AL and SL learning dynamics models are developed and used to illustrate the different implications of policy change in a standard New Keynesian model. We anticipate that the central question going forward will be how best to combine the two approaches when modeling adaptation to structural change.

Physics Models and Machine Learning in Molecular Dynamics

Abstract. Tackling the complexity of particle interactions, this investigation introduces a unified computational methodology employing Moldy, Gnuplot, and Visual Molecular Dynamics (VMD). Inspired by the n-body problem's persistent intrigue, specifically the three-body dynamics within molecular systems, we leverage Molecular Dynamics (MD) simulations to forecast and scrutinize the nuanced behavior of atomic and molecular entities. Moldy, a flexible MD software, facilitated the simulation of particle trajectories and interactions across diverse scenarios, with meticulous documentation of the resulting data. For visual analytics and insight extraction, Gnuplot crafted detailed plots depicting kinetic and potential energies alongside other thermodynamic metrics over temporal scales. The integration of VMD culminated in our analysis by vividly portraying the molecular motions, enhancing comprehension of emergent patterns. This study not only endorses Moldy's precision in MD simulations but also highlights the potent alliance among simulation, analysis, and visualization in elucidating intricate particle interactions. In summary, our work underscores the efficacy of the Moldy-Gnuplot-VMD triad as a formidable resource for scientists engaged in molecular motion prediction and analysis, paving fresh paths in computational physics and chemistry research.

Unmanned combat aerial vehicle path planning in complex environment using multi-strategy sparrow search algorithm with double-layer coding

Unmanned combat aerial vehicles (UCAV) path planning in complex environments demands a substantial number of path points to determine feasible paths. Establishing an effective flight path for UCAVs requires numerous path points to account for fuel constraints, artillery threats, and radar avoidance. This increase in path points raises the dimensionality of the problem, which in turn degrades algorithm performance. To mitigate this issue, a double-layer coding (DLC) model is utilized to remove redundant path points, consequently lowering computational complexity and operational difficulties. Meanwhile, this paper introduces a novel enhanced sparrow search algorithm (MESSA) based on multi-strategy for UCAV path planning. The MESSA incorporates a novel dynamic fitness regulation learning strategy (DFRL), a random differential learning strategy (RDL), an elite example equilibrium learning strategy (EEEL), a dynamic elimination and regeneration strategy based on the elite example (DERE), and quadratic interpolation (QI). Furthermore, MESSA is compared against 11 state-of-the-art algorithms, demonstrating exceptional optimization performance and robustness. Additionally, the combination of MESSA with the DLC model (DLC-MESSA) is applied to solve the UCAV path planning problem. The experimental results from five complex environments indicate that DLC-MESSA outperforms other algorithms in 80% of the cases by achieving the lowest average cost, thereby demonstrating its superior robustness and computational efficiency.

Robot Cognitive Learning by Considering Physical Properties

Humans achieve cognitive development through continuous interaction with their environment, enhancing both perception and behavior. However, current robots lack the capacity for human-like action and evolution, posing a bottleneck to improving robotic intelligence. Existing research predominantly models robots as one-way, static mappings from observations to actions, neglecting the dynamic processes of perception and behavior. This paper introduces a novel approach to robot cognitive learning by considering physical properties. We propose a theoretical framework wherein a robot is conceptualized as a three-body physical system comprising a perception-body (P-body), a cognition-body (C-body), and a behavior-body (B-body). Each body engages in physical dynamics and operates within a closed-loop interaction. Significantly, three crucial interactions connect these bodies. The C-body relies on the P-body’s extracted states and reciprocally offers long-term rewards, optimizing the P-body’s perception policy. In addition, the C-body directs the B-body’s actions through sub-goals, and subsequent P-body-derived states facilitate the C-body’s cognition dynamics learning. At last, the B-body would follow the sub-goal generated by the C-body and perform actions conditioned on the perceptive state from the P-body, which leads to the next interactive step. These interactions foster the joint evolution of each body, culminating in optimal design. To validate our approach, we employ a navigation task using a four-legged robot, D’Kitty, equipped with a movable global camera. Navigational prowess demands intricate coordination of sensing, planning, and D’Kitty’s motion. Leveraging our framework yields superior task performance compared with conventional methodologies. In conclusion, this paper establishes a paradigm shift in robot cognitive learning by integrating physical interactions across the P-body, C-body, and B-body, while considering physical properties. Our framework’s successful application to a navigation task underscores its efficacy in enhancing robotic intelligence.

Automated Diagnosis of Knee Osteoarthritis Using ResNet101 on a DEEP:PHI: Leveraging a No-Code AI Platform for Efficient and Accurate Medical Image Analysis.

Background: Knee osteoarthritis (OA) is a prevalent degenerative joint disease significantly impacting global health. Early and accurate diagnosis is crucial for effective management, but traditional methods often rely on subjective assessments. This study evaluates the efficacy of a deep learning model implemented through a no-code AI platform for diagnosing and grading knee OA from plain radiographs. Methods: We utilized the Osteoarthritis Initiative (OAI) dataset, comprising knee X-ray data from 1526 patients. The data were split into training (47.0%), validation (26.5%), and test (26.5%) sets. We employed a ResNet101 model on the DEEP:PHI no-code AI platform for image analysis. The model was trained to classify knee OA into five grades (0-4) based on the Kellgren-Lawrence scale. Results: Our AI model demonstrated high accuracy in distinguishing between different OA grades, with particular strength in early-stage detection. The model achieved optimal performance at 20 epochs, suggesting efficient learning dynamics. Grad-CAM visualizations were used to enhance the interpretability of the model's decision-making process. Conclusions: This study demonstrates the potential of AI, implemented through a no-code platform, to accurately diagnose and grade knee OA from radiographs. The use of a no-code AI platform such as DEEP:PHI represents a step towards democratizing AI in healthcare, enabling the rapid development and deployment of sophisticated medical AI applications without extensive coding expertise. This approach could significantly enhance the early detection and management of knee OA, potentially improving patient outcomes and streamlining clinical workflows.

On the shortcut learning in multilingual neural machine translation

In this study, we revisit the commonly-cited off-target issue in multilingual neural machine translation (MNMT). By carefully designing experiments on different MNMT scenarios and models, we attribute the off-target issue to the overfitting of the shortcuts of (non-centric, centric) language mappings. Specifically, the learned shortcuts biases MNMT to mistakenly translate non-centric languages into the centric language instead of the expected non-centric language for zero-shot translation. Analyses on learning dynamics show that the shortcut learning generally occurs in the later stage of model training, and multilingual pretraining accelerates and aggravates the shortcut learning. Based on these observations, we propose a simple and effective training strategy to eliminate the shortcuts in MNMT models by leveraging the forgetting nature of model training. The only difference from the standard training is that we remove the training instances that may induce the shortcut learning in the later stage of model training. Without introducing any additional data and computational costs, our approach can consistently and significantly improve the zero-shot translation performance by alleviating the shortcut learning for different MNMT models and benchmarks.