Learning Paradigm Research Articles

Semi-supervised few-shot class-incremental learning based on dynamic topology evolution

Incremental learning of new classes is crucial to developing real-world artificial intelligence systems. However, in some cases, we can only use a limited number of images to train the incremental model due to the difficulty of obtaining them. Few-shot class-incremental learning (FSCIL) is a new paradigm of incremental learning, which continually learns new concepts from a few examples. This setup can quickly lead to catastrophic forgetting and overfitting problems, severely affecting model performance. To address these problems, we propose a dynamic topological evolution (DTE) model that unifies knowledge of old and new categories. In this work, we focus on a novel semi-supervised few-shot class-incremental learning (SSFSCIL) to incrementally learn new categories with a limited number of labeled samples and a large number of unlabeled samples. In detail, we use the Voronoi polygon criterion to divide the feature space and construct a Hebbian graph to represent topological relationships between classes. Then, the corresponding k-nearest neighbor subgraph is constructed for the new incremental task, and label propagation of graph nodes is carried out on the unlabeled set to generate the corresponding pseudo labels. Finally, the new category knowledge is evolved into the actual category knowledge by integrating the Hebbian graph and the k-nearest neighbor subgraph obtained in the new incremental task. Extensive experiments on the mini-ImageNet and CIFAR100 datasets show that the accuracy of the DTE outperforms the baseline and improves by 1.41% and 1.13% on the last incremental task. Furthermore, our methodology achieved a 35.6% time improvement compared to full-supervised methods on the CIFAR100 dataset. In addition to the experiments on the two datasets mentioned above, the validation of our proposed DTE model extends to the engineering application domain of the RESICS-45 aircraft image scene classification dataset.

Federated 3D multi-organ segmentation with partially labeled and unlabeled data.

This paper considers a new problem setting for multi-organ segmentation based on the following observations. In reality, (1) collecting a large-scale dataset from various institutes is usually impeded due to privacy issues; (2) many images are not labeled since the slice-by-slice annotation is costly; and (3) datasets may exhibit inconsistent, partial annotations across different institutes. Learning a federated model from these distributed, partially labeled, and unlabeled samples is an unexplored problem. To simulate this multi-organ segmentation problem, several distributed clients and a central server are maintained. The central server coordinates with clients to learn a global model using distributed private datasets, which comprise a small part of partially labeled images and a large part of unlabeled images. To address this problem, a practical framework that unifies partially supervised learning (PSL), semi-supervised learning (SSL), and federated learning (FL) paradigms with PSL, SSL, and FL modules is proposed. The PSL module manages to learn from partially labeled samples. The SSL module extracts valuable information from unlabeled data. Besides, the FL module aggregates local information from distributed clients to generate a global statistical model. With the collaboration of three modules, the presented scheme could take advantage of these distributed imperfect datasets to train a generalizable model. The proposed method was extensively evaluated with multiple abdominal CT datasets, achieving an average result of 84.83% in Dice and 41.62mm in 95HD for multi-organ (liver, spleen, and stomach) segmentation. Moreover, its efficacy in transfer learning further demonstrated its good generalization ability for downstream segmentation tasks. This study considers a novel problem of multi-organ segmentation, which aims to develop a generalizable model using distributed, partially labeled, and unlabeled CT images. A practical framework is presented, which, through extensive validation, has proved to be an effective solution, demonstrating strong potential in addressing this challenging problem.

Relevance acquisition through motivational incentives: Modeling the time-course of associative learning and the role of visual features

Abstract Motivational relevance associated with symbolic stimuli impacts both neural and behavioral responses, similar to visual stimuli with inherent emotional valence. However, the specific effects of associated relevance on early sensory stages and lexico-semantic processing of these stimuli remain unclear, particularly considering the role of low-level visual features in relevance acquisition. To address these issues, we employed an associative learning paradigm in which we manipulated visual features, but not the stimuli themselves. The study (N = 48) included a learning phase, where pseudowords were associated with either gain, loss, or neutral outcomes. This was followed by a test phase the next day, involving an old/new decision task, in which stimuli were presented in either the same or a different font. During both phases, pupil responses and event-related brain potentials (P1, Early Posterior Negativity (EPN), Late Positive Complex (LPC), P3) were measured. Stronger pupil responses and increased neural activation in early visual encoding (P1) and lexico-semantic processing (EPN) were observed during relevance acquisition, particularly for loss associations. After relevance acquisition, the most substantial effect on modulating lexico-semantic processing was observed for gain associations, as evidenced by both behavioral responses and neural activity. During the test phase, exposure to incongruent visual features of the stimuli influenced the same processes that were observed during relevance acquisition. Notably, these effects of visual feature congruence were independent of those of associated motivational relevance. These results highlight the dynamic nature of motivational relevance effects, revealing differential effects observed during acquisition and the test phase, as well as between earlier perceptual processing and later neural and behavioral responses.

Multi-view Self-supervised Learning and Multi-scale Feature Fusion for Automatic Speech Recognition

To address the challenges of the poor representation capability and low data utilization rate of end-to-end speech recognition models in deep learning, this study proposes an end-to-end speech recognition model based on multi-scale feature fusion and multi-view self-supervised learning (MM-ASR). It adopts a multi-task learning paradigm for training. The proposed method emphasizes the importance of inter-layer information within shared encoders, aiming to enhance the model’s characterization capability via the multi-scale feature fusion module. Moreover, we apply multi-view self-supervised learning to effectively exploit data information. Our approach is rigorously evaluated on the Aishell-1 dataset and further validated its effectiveness on the English corpus WSJ. The experimental results demonstrate a noteworthy 4.6%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\%$$\\end{document} reduction in character error rate, indicating significantly improved speech recognition performance . These findings showcase the effectiveness and potential of our proposed MM-ASR model for end-to-end speech recognition tasks.

CSRNet: Focusing on critical points for depth completion

Depth completion is an effective method for generating dense depth maps from sparse ones. In recent studies, the majority of points, which we call standard points, often exhibit sub-optimal performance. This issue arises from the need to fit only very few points, termed as challenging points, which consist of noises and regions with discontinuous depth in the ground-truth. On the other hand, traditional evaluations can not recognize this situation, and are dominated by these limited challenging points, whose performance improvements may not significantly benefit related tasks. In contrast, standard points, which are critical for these tasks, are not effectively measured. This discrepancy highlights the need for a more targeted approach and evaluation method for depth completion. In order to solve the above problems, we propose a standard-point-enhancing learning paradigm. This paradigm aims to improve the performance on standard points, which consists of a Cascaded Segmentation-to-Regression Networks (CSRNet) and a Mining L1 loss. CSRNet includes two branches: DSNet and DRNet. DSNet uses segmentation to generate a coarse depth map, providing challenging-point-insensitive information. DRNet adopts a coarse-to-fine approach to learn the residual depth map between the coarse depth map and the ground-truth depth map. In addition, our Mining L1 loss leverages the segmentation results to filter out potential challenging points. This approach allows the network to concentrate more effectively on standard points. Lastly, we introduce the Minimum Error (ME) Curves as a new way to measure the performance of predicted depth maps in a flexible and comprehensive manner, irrespective of whether the points are standard or challenging. Experimental results on the KITTI and NYUDv2 datasets show that our approach significantly improves accuracy on the majority of points.

Professionalisation for inclusive mathematics—teacher education programs and changes in pre-service teachers’ beliefs and self-efficacy

Traditionally, in many countries there exist different teacher education programs for special education and for regular or inclusive education, not at all with similar underlying paradigms for teaching and learning or similar definitions and understanding of the term inclusion. In this context, one central question is how to define inclusive mathematics education, how to address the manifold aspects, and how to reduce discrimination and marginalization. On the one hand, teacher education programs might focus on diverse groups and specific students, like students with special needs and integrate important questions of special education. On the other hand, teacher education programs might take explicitly a broader perspective, considering the mathematical learning of all students and not taking a specific focus. In this contribution we present research results of the project ProViel (‘professionalisation for diversity’) to discuss how a teacher education program can be designed to address both with the underlying paradigm of making mathematics accessible for all students, and what role pre-service teachers’ beliefs as well as their self-efficacy might play. The project aimed at subject-specific concept development and research concerning teacher education for inclusive mathematics on the primary level, considering different points in time within the whole education program. Quantitative data of pre-service teachers’ beliefs and self-efficacy have been analysed, while they participated in different university courses and a practical phase at school. Key findings of this study are, that pre-service teachers’ beliefs about student achievement and their self-efficacy for inclusive mathematics teaching changed during the teacher education program.

Pengaruh Metode Snowball Throwing terhadap Hasil Belajar Pendidikan Agama Islam Kelas 5 SD Pembangunan Laboratorium UNP

This research was conducted because there are still students who experience difficulties in achieving good learning achievements in Islamic Religious Education subjects. The aim of this research is to evaluate the effect of the Snowball Throwing method in the 5th grade at UNP Laboratory Development Elementary School. The type of research is quasi-experimental. There are 36 students in class 5, each with 18 students in class 5A as the control class and 18 students in class 5B in the experimental class. Saturated sampling technique was used in the sampling process. Preliminary and final ability tests were used to collect research data. The results showed that the average score for the experimental class was 87.22. The learning objectives given in the form of a final exam are superior to the control class with a class average of 70.55. We can conclude that the learning outcomes of 5th grade elementary school students are influenced by the Snowball Throwing learning paradigm.

A sketch recognition method based on bi-modal model using cooperative learning paradigm

A sketch recognition method based on bi-modal model using cooperative learning paradigm

StudyNotion - An Edtech Website

In an era defined by rapid technological advancement, education stands as a crucial arena ripe for transformation. Our Edtech website emerges as a beacon of innovation, aiming to revolutionize traditional learning paradigms through the seamless integration of cutting-edge technology. With a multifaceted approach, our platform offers a comprehensive suite of tools and resources tailored to meet the diverse needs of students, educators, and institutions alike. At the heart of our platform lies a commitment to accessibility and inclusivity, ensuring that learning opportunities transcend geographical boundaries and socioeconomic barriers. Through interactive modules, engaging content, and personalized learning pathways, we empower learners to cultivate their intellectual curiosity and unlock their full potential. For educators, our platform serves as a dynamic hub for professional development, fostering collaboration, and knowledge sharing within a vibrant community of practice. From virtual classrooms to collaborative project management tools, we provide the scaffolding necessary to facilitate effective teaching in the digital age.Furthermore, our platform harnesses the power of data analytics and artificial intelligence to deliver actionable insights and personalized recommendations, facilitating data-driven decision-making and enhancing learning outcomes. In essence, our Edtech website represents more than just a digital learning platform; it embodies a vision for the future of education—one where technology serves as a catalyst for empowerment, innovation, and lifelong learning. Join us on this transformative journey as we redefine the possibilities of education in the 21st century and beyond. This abstract succinctly captures the essence of the Edtech website, highlighting its goals, features, and the transformative potential it holds for education.The abstract for the Edtech Website project encapsulates the endeavor to create an innovative educational platform tailored for diverse learners. This project aims to develop a user-friendly website offering interactive learning resources, personalized study tools, and collaborative features. Key objectives include providing seamless navigation, curating diverse educational content, implementing adaptive learning algorithms, and fostering real-time collaboration among students and educators. Emphasizing scalability, compatibility, and security, the website seeks to empower learners globally with accessible, engaging, and personalized learning experiences, ultimately fostering academic excellence and lifelong learning opportunities.

Conditional deletion of neurexin-2 impaired behavioral flexibility to alterations in action–outcome contingency

Neurexins (Nrxns) are critical for synapse organization and their mutations have been documented in autism spectrum disorder, schizophrenia, and epilepsy. We recently reported that conditional deletion of Nrxn2, under the control of Emx1Cre promoter, predominately expressed in the neocortex and hippocampus (Emx1-Nrxn2 cKO mice) induced stereotyped patterns of behavior in mice, suggesting behavioral inflexibility. In this study, we investigated the effects of Nrxn2 deletion through two different conditional approaches targeting presynaptic cortical neurons projecting to dorsomedial striatum on the flexibility between goal-directed and habitual actions in response to devaluation of action–outcome (A–O) contingencies in an instrumental learning paradigm or upon reversal of A–O contingencies in a water T-maze paradigm. Nrxn2 deletion through both the conditional approaches induced an inability of mice to discriminate between goal-directed and habitual action strategies in their response to devaluation of A–O contingency. Emx1-Nrxn2 cKO mice exhibited reversal learning deficits, indicating their inability to adopt new action strategies. Overall, our studies showed that Nrxn2 deletion through two distinct conditional deletion approaches impaired flexibility in response to alterations in A–O contingencies. These investigations can lay the foundation for identification of novel genetic factors underlying behavioral inflexibility.

Robust Multi-Agent Communication With Graph Information Bottleneck Optimization.

Recent research on multi-agent reinforcement learning (MARL) has shown that action coordination of multi-agents can be significantly enhanced by introducing communication learning mechanisms. Meanwhile, graph neural network (GNN) provides a promising paradigm for communication learning of MARL. Under this paradigm, agents and communication channels can be regarded as nodes and edges in the graph, and agents can aggregate information from neighboring agents through GNN. However, this GNN-based communication paradigm is susceptible to adversarial attacks and noise perturbations, and how to achieve robust communication learning under perturbations has been largely neglected. To this end, this paper explores this problem and introduces a robust communication learning mechanism with graph information bottleneck optimization, which can optimally realize the robustness and effectiveness of communication learning. We introduce two information-theoretic regularizers to learn the minimal sufficient message representation for multi-agent communication. The regularizers aim at maximizing the mutual information (MI) between the message representation and action selection while minimizing the MI between the agent feature and message representation. Besides, we present a MARL framework that can integrate the proposed communication mechanism with existing value decomposition methods. Experimental results demonstrate that the proposed method is more robust and efficient than state-of-the-art GNN-based MARL methods.

REINFORCEMENT LEARNING: FRAMEWORK, APPLICATIONS AND CHALLENGES

Artificial Intelligence is the science of developing intelligent machines that can simulate human intelligence and perform complex tasks. Reinforcement learning (RL) is the trending interdisciplinary learning paradigm of machine learning and optimal control that mirrors the hit and trial learning mechanism of humans .The self-learning agents take actions in a dynamic environment, get rewarded or punished and use this feedback to learn optimal decision-making to maximize cumulative rewards over time. This objective of this paper is to present an insight into promising future technology. The introduction to the technology is followed by the key concepts that constitute the reinforcement framework, the different models of this paradigm, the applications that span a wide arena and the challenges facing this ever-evolving strategy.

The new paradigm in machine learning - foundation models, large language models and beyond: a primer for physicians.

Foundation machine learning models are deep learning models capable of performing many different tasks using different data modalities such as text, audio, images and video. They represent a major shift from traditional task-specific machine learning prediction models. Large language models (LLM), brought to wide public prominence in the form of ChatGPT, are text-based foundational models that have the potential to transform medicine by enabling automation of a range of tasks, including writing discharge summaries, answering patients questions and assisting in clinical decision-making. However, such models are not without risk and can potentially cause harm if their development, evaluation and use are devoid of proper scrutiny. This narrative review describes the different types of LLM, their emerging applications and potential limitations and bias and likely future translation into clinical practice.

Decoding trust: a reinforcement learning perspective

Abstract Behavioral experiments on the trust game have shown that trust and trustworthiness are commonly seen among human beings, contradicting the prediction by assuming Homo economicus in orthodox Economics. This means some mechanism must be at work that favors their emergence. Most previous explanations, however, need to resort to some exogenous factors based upon imitative learning, a simple version of social learning. Here, we turn to the paradigm of reinforcement learning, where individuals revise their strategies by evaluating the long-term return through accumulated experience. Specifically, we investigate the trust game with the Q-learning algorithm, where each participant is associated with two evolving Q-tables that guide one’s decision-making as trustor and trustee, respectively. In the pairwise scenario, we reveal that high levels of trust and trustworthiness emerge when individuals appreciate both their historical experience and returns in the future. Mechanistically, the evolution of the Q-tables shows a crossover that resembles human psychological changes. We also provide the phase diagram for the game parameters, where the boundary analysis is conducted. These findings are robust when the scenario is extended to a latticed population. Our results thus provide a natural explanation for the emergence of trust and trustworthiness, and indicate that the long-ignored endogenous factors alone are sufficient to drive. More importantly, the proposed paradigm shows the potential to decipher many puzzles in human behaviors.

REINFORCEMENT LEARNING: FRAMEWORK, APPLICATIONS AND CHALLENGES

Artificial Intelligence is the science of developing intelligent machines that can simulate human intelligence and perform complex tasks. Reinforcement learning (RL) is the trending interdisciplinary learning paradigm of machine learning and optimal control that mirrors the hit and trial learning mechanism of humans .The self-learning agents take actions in a dynamic environment, get rewarded or punished and use this feedback to learn optimal decision-making to maximize cumulative rewards over time. This objective of this paper is to present an insight into promising future technology. The introduction to the technology is followed by the key concepts that constitute the reinforcement framework, the different models of this paradigm, the applications that span a wide arena and the challenges facing this ever-evolving strategy.

Diverse memory paradigms in Drosophila reveal diverse neural mechanisms.

In this review, we aggregated the different types of learning and memory paradigms developed in adult Drosophila and attempted to assess the similarities and differences in the neural mechanisms supporting diverse types of memory. The simplest association memory assays are conditioning paradigms (olfactory, visual, and gustatory). A great deal of work has been done on these memories, revealing hundreds of genes and neural circuits supporting this memory. Variations of conditioning assays (reversal learning, trace conditioning, latent inhibition, and extinction) also reveal interesting memory mechanisms, whereas mechanisms supporting spatial memory (thermal maze, orientation memory, and heat box) and the conditioned suppression of innate behaviors (phototaxis, negative geotaxis, anemotaxis, and locomotion) remain largely unexplored. In recent years, there has been an increased interest in multisensory and multicomponent memories (context-dependent and cross-modal memory) and higher-order memory (sensory preconditioning and second-order conditioning). Some of this work has revealed how the intricate mushroom body (MB) neural circuitry can support more complex memories. Finally, the most complex memories are arguably those involving social memory: courtship conditioning and social learning (mate-copying and egg-laying behaviors). Currently, very little is known about the mechanisms supporting social memories. Overall, the MBs are important for association memories of multiple sensory modalities and multisensory integration, whereas the central complex is important for place, orientation, and navigation memories. Interestingly, several different types of memory appear to use similar or variants of the olfactory conditioning neural circuitry, which are repurposed in different ways.

Upaya Meningkatkan Keterampilan Kolaborasi dan Prestasi Belajar Matematika pada Materi Pecahan Menggunakan Model Problem Based Learning Berbantu Media Konkret Kelas V SD Negeri 2 Sangkanayu

The objective of this study is to improve the collaboration skills of fifth-grade students and mathematics learning achievement in the fraction topic at SD Negeri 2 Sangkanayu by utilizing the problem-based learning (PBL) paradigm with concrete media. This study used a two-cycle method for classroom action research (CAR). Eachcycle consisted of four steps: planning, implementation, assessment, and reflection. Nineteen students from thefifth grade of SD Negeri 2 Sangkanayu became the research subjects. Tests with assessment questions and non-test surveys about collaboration skills are examples of data instrument strategies. The research findings indicate that students' ability to collaborate with others and perform academically has improved. The average learning outcome in cycle I was 60.65 with a learning completion rate of 65.7%, while the learning achievement results in cycle II obtained an average of 83.68 with a learning completion rate of 86.8%. The average percentage of collaboration skills increased 5.1% from 75.85% in cycle I to 81% in cycle II.

Sequential Label Enhancement.

Label distribution learning (LDL) is a novel machine learning paradigm for solving ambiguous tasks, where the degree to which each label describing the instance is ambiguous. However, obtaining the label distribution is high cost and the description degree is difficult to quantify. Most existing research works focus on designing an objective function to obtain the whole description degrees at once but seldom care about the sequentiality in the process of recovering the label distribution. In this article, we formulate the label distribution recovering task as a sequential decision process called sequential label enhancement (Seq_LE), which is more consistent with the process of annotating the label distribution in human brains. Specifically, the discrete label and its description degree are serially mapped by the reinforcement learning (RL) agent. Besides, we carefully design a joint reward function to drive the agent to fully learn the optimal decision policy. Extensive experiments on 16 LDL datasets are conducted under various evaluation metrics. The experimental results demonstrate convincingly that the proposed sequential label enhancement (LE) leads to better performance over the state-of-the-art methods.

Learning With Noisy Labels Over Imbalanced Subpopulations.

Learning with noisy labels (LNL) has attracted significant attention from the research community. Many recent LNL methods rely on the assumption that clean samples tend to have a "small loss." However, this assumption often fails to generalize to some real-world cases with imbalanced subpopulations, that is, training subpopulations that vary in sample size or recognition difficulty. Therefore, recent LNL methods face the risk of misclassifying those "informative" samples (e.g., hard samples or samples in the tail subpopulations) into noisy samples, leading to poor generalization performance. To address this issue, we propose a novel LNL method to deal with noisy labels and imbalanced subpopulations simultaneously. It first leverages sample correlation to estimate samples' clean probabilities for label correction and then utilizes corrected labels for distributionally robust optimization (DRO) to further improve the robustness. Specifically, in contrast to previous works using classification loss as the selection criterion, we introduce a feature-based metric that takes the sample correlation into account for estimating samples' clean probabilities. Then, we refurbish the noisy labels using the estimated clean probabilities and the pseudo-labels from the model's predictions. With refurbished labels, we use DRO to train the model to be robust to subpopulation imbalance. Extensive experiments on a wide range of benchmarks demonstrate that our technique can consistently improve state-of-the-art (SOTA) robust learning paradigms against noisy labels, especially when encountering imbalanced subpopulations. We provide our code in https://github.com/chenmc1996/LNL-IS.

Deep Class-Incremental Learning From Decentralized Data.

In this article, we focus on a new and challenging decentralized machine learning paradigm in which there are continuous inflows of data to be addressed and the data are stored in multiple repositories. We initiate the study of data-decentralized class-incremental learning (DCIL) by making the following contributions. First, we formulate the DCIL problem and develop the experimental protocol. Second, we introduce a paradigm to create a basic decentralized counterpart of typical (centralized) CIL approaches, and as a result, establish a benchmark for the DCIL study. Third, we further propose a decentralized composite knowledge incremental distillation (DCID) framework to transfer knowledge from historical models and multiple local sites to the general model continually. DCID consists of three main components, namely, local CIL, collaborated knowledge distillation (KD) among local models, and aggregated KD from local models to the general one. We comprehensively investigate our DCID framework by using a different implementation of the three components. Extensive experimental results demonstrate the effectiveness of our DCID framework. The source code of the baseline methods and the proposed DCIL is available at https://github.com/Vision-Intelligence-and-Robots-Group/DCIL.