Learning Task Research Articles

Bypassing Striatal Learning Mechanisms Using Delayed Feedback to Circumvent Learning Deficits in Traumatic Brain Injury.

Feedback facilitates learning by guiding and modifying behaviors through an action-outcome contingency. As the majority of existing studies have focused on the immediate presentation of feedback, the impact of delayed feedback on learning is understudied. Prior work demonstrated that learning from immediate and delayed feedback employed distinct brain regions in healthy individuals, and compared to healthy individuals, individuals with traumatic brain injury (TBI) are impaired in learning from immediate feedback. The goal of the current investigation was to assess the effects of delayed vs immediate feedback on learning in individuals with TBI and examine brain networks associated with delayed and immediate feedback processing. Nonprofit research organization. Twenty-eight individuals with moderate-to-severe TBI. Participants completed a paired-associate word learning task while undergoing magnetic resonance imaging. During the task, feedback was presented either immediately, after a delay, or not at all (control condition). Learning performance accuracy, confidence ratings, post-task questionnaire, and blood oxygen level-dependent signal. Behavioral data showed that delayed feedback resulted in better learning performance than immediate feedback and no feedback. In addition, participants reported higher confidence in their performance during delayed feedback trials. During delayed vs immediate feedback processing, greater activation was observed in the superior parietal and angular gyrus. Activation in these areas has been previously associated with successful retrieval and greater memory confidence. The observed results might be explained by delayed feedback processing circumventing the striatal dopaminergic regions responsible for learning from immediate feedback that are impaired in TBI. In addition, delayed feedback evokes less of an affective reaction than immediate feedback, which likely benefited memory performance. Indeed, compared to delayed feedback, positive or negative immediate feedback was more likely to be rated as rewarding or punishing, respectively. The findings have significant implications for TBI rehabilitation and suggest that delaying feedback during rehabilitation might recruit brain regions that lead to better functional outcomes.

Few-shot learning with long-tailed labels

Few-Shot Learning (FSL) is a challenging classification task in machine learning, and it aims to recognize unseen examples of new classes with only a few labeled reference examples (i.e., the support set). The training phase of FSL typically requires a large amount of labeled examples (i.e., the base set) to effectively learn transferable knowledge, but it is usually difficult to obtain sufficient data annotation in practical applications. Existing semi-supervised FSL approaches can learn generalizable representations from partly labeled data, yet they do not sufficiently consider the real distribution of those labeled data. In this paper, we propose a new problem setting termed Few-Shot Learning with Long-Tailed Labels (FSL-LTL) to further consider a more practical semi-supervised scenario where the labeled examples are long-tailed. To effectively address this new problem, we build a novel two-stage training framework dubbed Reweighted Contrastive Embedding (RCE). In the first stage of RCE, we adopt the popular contrastive learning framework to pre-train a reliable network in a self-supervised manner. In the second stage, we integrate the semi-supervised empirical risk into a Weighted Random Sampling (WRS) strategy to fine-tune the pre-trained backbone with the aid of a consistency regularization. Experimental results demonstrate the feasibility of the proposed FSL-LTL problem setting and the superiority of our new RCE method over existing FSL approaches and semi-supervised learning methods. These results also suggest that the RCE approach is a promising solution for addressing the new FSL-LTL problem.

Inductive biases of neural network modularity in spatial navigation.

The brain may have evolved a modular architecture for daily tasks, with circuits featuring functionally specialized modules that match the task structure. We hypothesize that this architecture enables better learning and generalization than architectures with less specialized modules. To test this, we trained reinforcement learning agents with various neural architectures on a naturalistic navigation task. We found that the modular agent, with an architecture that segregates computations of state representation, value, and action into specialized modules, achieved better learning and generalization. Its learned state representation combines prediction and observation, weighted by their relative uncertainty, akin to recursive Bayesian estimation. This agent's behavior also resembles macaques' behavior more closely. Our results shed light on the possible rationale for the brain's modularity and suggest that artificial systems can use this insight from neuroscience to improve learning and generalization in natural tasks.

Four Cases of Children With Phonological Impairment and Precocious Vocabulary: Making Sense of a Clinical Conundrum.

The purpose of this study was to describe the speech production, speech perception, vocabulary, and word learning abilities of lexically precocious 4-year-old children with phonological impairment, in an effort to better understand the underlying nature of phonological impairment in children. Using a case series approach, we identified four children with phonological impairment and precocious vocabulary abilities. Each child completed routine speech production and vocabulary assessments, as well as experimental speech perception and word learning tasks. The results from these tasks were then used to create profiles of each child's individual strengths and needs across the abilities assessed. Although all four children presented with phonological impairment and lexically precocious receptive and expressive vocabulary, they differed in their specific speech errors. One child presented with phonological speech errors only, while the other three children presented with an interdental lisp alongside their phonological errors. Three children presented with average speech perception abilities, and one child presented with poorer speech perception. The same three children also showed some learning of novel nonwords 1 week post-initial exposure, while the other child showed no evidence of word learning 1 week post-initial exposure. The clinical profiles of lexically precocious children with phonological impairment offered different insights into the nature of the problem. Although one child appeared to present with underspecified underlying representations of words, the other three children appeared to present with well-specified underlying representations. Of the three children with well-specified underlying representations, two appeared to have difficulty abstracting particular rules of the ambient phonological system. Further research is needed to improve our understanding of the underlying nature of phonological impairment. https://doi.org/10.23641/asha.26307640.

Amygdala and cortex relationships during learning of a sensory discrimination task.

During learning of a sensory discrimination task, cortical and sub-cortical regions display complex spatiotemporal dynamics. During learning, both amygdala and cortex link stimulus information to its appropriate association, for example a reward. In addition, both structures are also related to non-sensory parameters such as body movements and licking during the reward period. However, the emergence of cortico-amygdala relationships during learning is largely unknown. To study this, we combined wide-field cortical imaging with fiber photometry to simultaneously record cortico-amygdala population dynamics as male mice learn a whisker-dependent go/no-go task. We were able to simultaneously record neuronal populations from the posterior cortex and either the basolateral amygdala (BLA) or central medial amygdala (CEM). Prior to learning, somatosensory and associative cortex responded during sensation, while amygdala areas did not show significant responses. As mice became experts, amygdala responses emerged early during the sensation period, increasing in CEM, while decreasing in BLA. Interestingly, amygdala and cortical responses were associated with task-related body movement, displaying significant responses ∼200 ms before movement initiation which led to licking for the reward. A correlation analysis between cortex and amygdala revealed negative and positive correlation with BLA and CEM respectively, only in the expert case. These results imply that learning induces an involvement of cortex and amygdala which may aid to link sensory stimuli with appropriate associations.Significance Statement Levitan and Gilad study neuronal dynamics in cortex and amygdala as mice learn a sensory discrimination task. They find that amygdala areas display opposing responses only after the mice learn the task. In contrast cortical areas display responses both before and after learning. These results imply that learning induces an involvement of amygdala and cortical areas in order to optimally link sensory stimuli with appropriate associations.

Borderline personality disorder and learning: The influences of emotional state and social versus nonsocial feedback

BackgroundBorderline personality disorder (BPD) has been associated with decision-making deficits, yet such deficits may be context dependent, particularly emotional state and social context. Reinforcement learning models offer an avenue to pinpoint decision-making impairments. The current study used reinforcement learning models to examine whether feedback type (social vs. nonsocial) or emotional state (neutral vs. negative) influence the association between BPD and decision making. MethodAdults (N = 131) with a range of BPD symptoms completed a diagnostic interview and a computerized learning task after neutral and negative emotion inductions. We examined accuracy, learning rate, and exploration. ResultsWe conducted linear models to examine the association between BPD criteria, feedback type, and emotional state on learning parameters and learning accuracy. We found that the negative emotion condition was associated with greater exploration, particularly for those with elevated BPD features. Furthermore, elevated BPD features were associated with impaired accuracy when aiming to avoid loss. A 3-way interaction between BPD, emotion, and feedback indicated that, for people with higher BPD features, learning performance was further impaired when receiving social feedback in the negative emotion condition. LimitationsSeveral limitations warrant mention, including a relatively homogenous sample, possible co-occurring diagnoses, and methodological consideration with the learning task. ConclusionsThe present study underscored the link between BPD and learning impairments. Amplified learning alterations under negative social contexts have important implications for identifying optimal venues to teach new skills (of relevance to treatment) for those with BPD.

Prioritized adjustments in posture stabilization and adaptive reaching during neuromuscular fatigue of lower-limb muscles.

Neuromuscular fatigue (NMF) induces temporary reductions in muscle force production capacity, affecting various aspects of motor function. While studies have extensively explored NMF's impact on muscle activation patterns and postural stability, its influence on motor adaptation processes remains less understood. This paper investigates the effects of localized NMF on motor adaptation during upright stance, focusing on reaching tasks. Utilizing a force field perturbation paradigm, participants performed reaching movements while standing upright before and after inducing NMF in the ankle dorsiflexor muscles. Results revealed that despite maintained postural stability, participants in the NMF group exhibited larger movement errors during reaching tasks, suggesting impaired motor adaptation. This was evident in both initial and terminal phases of adaptation, indicating a disruption in learning processes rather than a decreased adaptation rate. Analysis of electromyography activation patterns highlighted distinct strategies between groups, with the NMF group showing altered activation of both fatigued and non-fatigued muscles. Additionally, differences in co-activation patterns suggested compensatory mechanisms to prioritize postural stability despite NMF-induced disruptions. These findings underscore the complex interplay between NMF, motor adaptation, and postural control, suggesting a potential role for central nervous system mechanisms in mediating adaptation processes. Understanding these mechanisms has implications for sports performance, rehabilitation, and motor skill acquisition, where NMF may impact the learning and retention of motor tasks. Further research is warranted to elucidate the transient or long-term effects of NMF on motor adaptation and its implications for motor rehabilitation interventions.

Examining the role of stimulus complexity in item and associative memory.

Episodic memory comprises memory for individual information units (item memory) and for the connections among them (associative memory). In two experiments using an object pair learning task, we examined the effect of visual stimulus complexity on memory encoding and retrieval mechanisms and on item and associative memory performance. Subjects encoded pairs of black monochrome object images (low complexity, LC condition) or color photographs of objects (high complexity, HC condition) via interactive imagery, and subsequently item and associative recognition were tested. In Experiment 1, event-related potentials (ERPs) revealed an enhanced frontal N2 during encoding and an enhanced late posterior negativity (LPN) during item recognition in the HC condition, suggesting that memory traces containing visually more complex objects elicited a stronger effort in reconstructing the past episode. Item memory was consistently superior in the HC compared to the LC condition. Associative memory was either statistically unaffected by complexity (Experiment 1) or improved (Experiment 2) in the HC condition, speaking against a tradeoff between resources allocated to item versus associative memory, and hence contradicting results of some prior studies. In Experiment 2, in both young and older adults, both item and associative memory benefitted from stimulus complexity, such that the magnitude of the age-related associative deficit was not influenced by stimulus complexity. Together, these results suggest that if familiar objects are presented in a form that exhibits a higher visual complexity, which may support semantic processing, complexity can benefit both item and associative memory. Stimulus properties that enhance item memory can scaffold associative memory in this situation.

Implicit and Explicit Sequence Learning in Adults With Developmental Language Disorder.

Developmental language disorder (DLD) is a neurodevelopmental disorder that impacts approximately 7% of the population and is characterized by unexplained deficits in expressive and/or receptive components of language. A common procedural learning task, serial reaction time (SRT), has been used to develop models of the basis of DLD. However, paradigms involve differing levels of implicit and explicit learning during this task, muddying interpretations of the data. Here, we tested adults with DLD on implicit and explicit SRT tasks to better understand implicit and explicit procedural learning in this population. We hypothesized that adults with DLD would demonstrate reduced learning on only the implicit SRT task, as alternate explicit neural mechanisms could lead to equivalent performance on the explicit task. Fifty participants (25 with DLD and 25 with typical language) completed implicit and explicit SRT tasks, measuring their ability to learn visually presented 10-element sequences. Group differences were evaluated on sequence learning, error rates, and explicit recall of the sequence after learning. Sequence learning was the same between the groups on both tasks. However, individuals with DLD showed increased errors and significantly worse recall of the explicitly learned sequence. Results suggest that sequence learning may be intact in this population, while aspects of explicit learning and motoric responses are impaired. Results are interpreted in light of a neurobiological model of DLD. https://doi.org/10.23641/asha.26210651.

Exploring different content creation and display methods for remote collaboration supported by eXtended reality: comparative analysis of distinct task scenarios

AbstractRemote collaboration using eXtended Reality (XR) has been explored to establish a common ground between physically distributed individuals. To achieve usable and impactful solutions, it is paramount to understand how different content creation and display methods contribute to the work effort of each member (remote and on-site). Additionally, explore how various task scenarios influence the collaborative process, specifically examining how activities with unique characteristics and complexities affect remote work. Hence, the question arises, ’How do different display and interaction methods impact the collaborative process within specific task domains?’ In this paper, two user studies with 30 distinct participants each are described, focusing on different content creation support (Laptop Computer; Video Wall & Keyboard; Interactive Projector) and display methods (Hand-Held Device (HHD); HHD & Articulated Support; Head-Mounted Display (HMD)) for remote and on-site collaborators respectively, during scenarios of remote guidance. Plus, different physical tasks were considered for each study (Lego pieces assembly; Tangram puzzle assembly; Maintenance procedures; Resource management in a map; Training activity in a laboratory classroom), to understand which method stands out according to the characteristics of the said tasks. We report the results obtained, suggesting that for the remote part of the collaborative process, using a Video Wall & Keyboard appears to be more suitable for tasks associated with maintenance procedures, learning, and training activities. As for the on-site part, using HMD was considered the better option for maintenance, and learning tasks, closely followed by HHD & Articulated Support.

Game-theoretic distributed approach for heterogeneous-cost task allocation with budget constraints

This paper investigates heterogeneous-cost task allocation with budget constraints (HCTAB), wherein heterogeneity is manifested through the varying capabilities and costs associated with different agents for task execution. In contrast with the centralized optimization-based method, the HCTAB problem is solved using a fully distributed framework, and a coalition formation game is introduced to provide a theoretical guarantee for this distributed framework. To solve the coalition formation game, a convergence-guaranteed log-linear learning algorithm based on heterogeneous cost is proposed. This algorithm incorporates two improvement strategies, namely a cooperative exchange strategy and a heterogeneous-cost log-linear learning strategy. These strategies are specifically designed to be compatible with the heterogeneous cost and budget constraints characteristic of the HCTAB problem. Through ablation experiments, we demonstrate the effectiveness of these two improvements. Finally, numerical results show that the proposed algorithm outperforms existing task allocation algorithms and learning algorithms in terms of solving the HCTAB problem.

Fertilizer and Crop Yield Prediction using Machine Learning

The agricultural sector is indispensable to feeding the growing global population, making efficient crop management and yield prediction imperative. Traditional farming practices often rely on subjective decision-making and generalized fertilizer application methods, leading to suboptimal resource utilization and yield outcomes. In this research, we introduce an innovative method Utilizing the capability the bunch of algorithms introduced for machine learning tasks to precise fertilizer recommendation and crop yield prediction. The developed system provides farmers with personalized fertilizer recommendations tailored to their specific soil and crop requirements, thereby minimizing waste and maximizing yield potential. Additionally, real-time monitoring and feedback mechanisms enable adaptive adjustments throughout the growing season, ensuring timely interventions to mitigate adverse outcomes and optimize productivity

Repetitive active and passive cognitive stimulations induce EEG changes in patients with Rett syndrome.

Despite being considered a rare disease, Rett syndrome is a leading cause of profound cognitive impairment in females. This study explores game-based cognitive stimulation to enhance attention during learning tasks, offering an alternative treatment perspective. Fifteen diagnosed Rett syndrome girls participated in four 24-minute sessions, including a 5-minute initial resting state recording. Primary indicators for analysis included relative power and spectral entropy. Significant findings indicated variations among conditions (resting state, active task, passive task) in response to stimulation. Notably, over four days, evolution occurred, characterized by decreasing delta power and increasing theta and beta power. Topographic maps confirmed these shifts, highlighting affected brain areas. Linear regression emphasized the most significant impact on the first day, with subsequent shifts towards higher frequencies, particularly during the resting state. By the fourth day, resting-state patterns resembled those during cognitive activities. Findings suggest cognitive stimulation induces substantial EEG spectral changes, potentially linked to cognitive enhancements in Rett syndrome. The shift towards higher frequency bands and increased spectral entropy align with enhanced brain activation during cognitive sessions, underscoring the potential of cognitive stimulation therapies and calling for further research to optimize abilities in individuals with Rett syndrome. Game-based cognitive stimulation induces substantial EEG changes in individuals with Rett syndrome, enhancing cognitive functions, notably attention during learning. This study conducts a distinctive examination, assessing the habituation paradigm through the combination of game-based cognitive stimulation and learning, providing valuable insights into enhancing attention in Rett syndrome. Impacting understanding of cognitive processes in Rett syndrome, this research reveals significant EEG variations during tasks, emphasizing the potential of cognitive stimulation for attention enhancement and the need for further research in tailored interventions.

Evolution and advancements in deep learning models for Natural Language Processing

This paper provides a comprehensive review of the evolution and advancements in deep learning models for Natural Language Processing (NLP). It explores the transition from statistical models to neural networks, highlighting the paradigm shift towards data-driven methodologies and the implications for NLP tasks. The emergence of neural network architectures, such as Recurrent Neural Networks (RNNs) and transformer-based models like BERT and GPT, has revolutionized language understanding and generation. Furthermore, the integration of deep learning in traditional NLP tasks, such as part-of-speech tagging and named entity recognition, has led to significant improvements in accuracy and efficiency. The paper also discusses the quantitative analysis of deep learning models, including performance metrics, computational efficiency, and mathematical modeling of language tasks. Case studies and applications, including sentiment analysis, machine translation, and automated content generation, exemplify the transformative impact of deep learning in NLP.

Parallel diffusion models promote high detail-fidelity photoacoustic microscopy in sparse sampling

Reconstructing sparsely sampled data is fundamental for achieving high spatiotemporal resolution photoacoustic microscopy (PAM) of microvascular morphology in vivo. Convolutional networks (CNN) and generative adversarial networks (GAN) have been introduced to high-speed PAM, but due to the use of upsampling in CNN-based networks to restore details and the instability in GAN training, they struggle to learn the entangled microvascular network structure and vascular texture features, resulting in only achieving low detail-fidelity imaging of microvascular. The diffusion models is richly sampled and can generate high-quality images, which is very helpful for the complex vascular features in PAM. Here, we propose an approach named parallel diffusion models (PDM) with parallel learning of Noise task and Image task, where the Noise task optimizes through variational lower bounds to generate microvascular structures that are visually realistic, and the Image task improves the fidelity of the generated microvascular details through image-based loss. With only 1.56% of fully sampled pixels from photoacoustic human oral data, PDM achieves an LPIPS of 0.199. Additionally, using PDM in high-speed 16x PAM prevents breathing artifacts and image distortion issues caused by low-speed sampling, reduces the standard deviation of the Row-wise Self-Correlation Coefficient, and maintains high image quality. It achieves high confidence in reconstructing detailed information from sparsely sampled data and will promote the application of reconstructed sparsely sampled data in realizing high spatiotemporal resolution PAM.

A Deep Learning-based Classification Model for Arabic News Tweets Using Bidirectional Long Short-Term Memory Networks

This research develops a classification model for Arabic news tweets using Bidirectional Long Short-Term Memory networks (BiLSTM). Tweets about Arabic news were gathered between August 2016 and August 2020 and divided into five categories. Custom Python scripts, Twitter API and the GetOldTweets3 Python library were used to collect the data. BiLSTM was used to train and test the model. The results indicated an average accuracy, precision, recall, and f1-score of 0.88, 0.92, 0.88, and 0.89, respectively. The results could have practical implications for Arabic machine learning and NLP tasks in research and practice.

Feedback and observational learning differ in effectiveness during category learning in early school aged children and adults.

When learning new categories, do children benefit from the same types of training as adults? We compared the effects of feedback-based training with observational training in young adults (ages 18-25) and early school aged children (ages 6-7) across two different multimodal category learning tasks: conjunctive rule based and information integration. We used multimodal stimuli that varied across a visual feature (rotation speed of the "planet" stimulus) and an auditory feature (pitch frequency of a pure tone stimulus). We found an interaction between age and training type for the rule-based category task, such that adults performed better in feedback training than in observational training, whereas training type had no significant effect on children's category learning performance. Overall adults performed better than children in learning both the rule based and information integration category structures. In information integration category learning, feedback versus observational training did not have a significant effect on either adults' or children's category learning. Computational modelling revealed that children defaulted to univariate rules in both tasks. The finding that children do not benefit from feedback training and can learn successfully via observational learning has implications for the design of educational interventions appropriate for children.

Implementing the analogous neural network using chaotic strange attractors

Machine learning studies need colossal power to process massive datasets and train neural networks to reach high accuracies, which have become gradually unsustainable. Limited by the von Neumann bottleneck, current computing architectures and methods fuel this high power consumption. Here, we present an analog computing method that harnesses chaotic nonlinear attractors to perform machine learning tasks with low power consumption. Inspired by neuromorphic computing, our model is a programmable, versatile, and generalized platform for machine learning tasks. Our mode provides exceptional performance in clustering by utilizing chaotic attractors’ nonlinear mapping and sensitivity to initial conditions. When deployed as a simple analog device, it only requires milliwatt-scale power levels while being on par with current machine learning techniques. We demonstrate low errors and high accuracies with our model for regression and classification-based learning tasks.

Unsupervised Imputation of Non-Ignorably Missing Data Using Importance-Weighted Autoencoders

Deep Learning (DL) methods have dramatically increased in popularity in recent years. While its initial success was demonstrated in the classification and manipulation of image data, there has been significant growth in the application of DL methods to problems in the biomedical sciences. However, the greater prevalence and complexity of missing data in biomedical datasets present significant challenges for DL methods. Here, we provide a formal treatment of missing data in the context of Variational Autoencoders (VAEs), a popular unsupervised DL architecture commonly used for dimension reduction, imputation, and learning latent representations of complex data. We propose a new VAE architecture, NIMIWAE, that is one of the first to flexibly account for both ignorable and non-ignorable patterns of missingness in input features at training time. Following training, samples can be drawn from the approximate posterior distribution of the missing data can be used for multiple imputation, facilitating downstream analyses on high dimensional incomplete datasets. We demonstrate through statistical simulation that our method outperforms existing approaches for unsupervised learning tasks and imputation accuracy. We conclude with a case study of an EHR dataset pertaining to 12,000 ICU patients containing a large number of diagnostic measurements and clinical outcomes, where many features are only partially observed.

Improving the Learning of Code Review Successive Tasks with Cross-Task Knowledge Distillation

Improving the Learning of Code Review Successive Tasks with Cross-Task Knowledge Distillation