Training Tasks Research Articles

An Inanimate Intracorporeal Anastomosis Model With Real-Time Force Feedback: An Initial Study.

Laparoscopic intestinal anastomosis requires specific technical skills and should be trained in a safe simulation environment before performing surgery in daily practice. However, anastomosis simulation training with objective feedback is not widely available. This study aimed to analyze a laparoscopic intestinal anastomosis training task that utilizes objective force, motion, and time measurements. With the feedback of laparoscopic experts, an artificial tissue reproducible intestinal anastomosis training task was designed and developed. Novices and experts performed the training task four times using two running suture techniques, with a multifilament braided suture and a barbed suture. The laparoscopic box trainer (Lapron box trainer, Amsterdam Skills Centre, Amsterdam, the Netherlands) provided objective force, motion, and time feedback. The mean values of the parameters were calculated and analyzed using the Mann-Whitney U test. A total of 212 intestinal anastomosis repetitions were performed by 35 novices and 18 experts from 14 European teaching hospitals. For the multifilament braided sutures, experts showed significant lower maximal impulses (19.80 versus 12.90Ns, P=0.004), shorter total path length (23,545mm versus 15,266mm, P≤0.001) and required less time to finish the task compared to novices (448s versus 297s, P≤0.001). Using the barbed sutures, experts used significantly lower maximal forces (2.93N versus 2.31N, P=0.032), had a shorter total path length (13,608mm versus 8551mm, P≤0.001), and needed less time to execute the training task compared to novices (253s versus 166s, P≤0.001). The development of a modular and reproducible laparoscopic intestinal anastomosis training task with established construct validity for force, motion, and time-based assessment of technical skills allows for repetitive training of advanced skills. These outcomes can now be utilized to assess translation of these skills into the operating room.

Impact of Virtual Reality Task-Oriented Training on Upper Extremity Motor Function in Children with Cerebral Palsy

This study aims to investigate the effect of a newly developed virtual reality task-oriented training (VR-TOT) video game on upper extremity fine motor function compared with conventional occupational therapy through leap motion in children with spastic hemiplegic cerebral palsy (CP). In this double-blind randomized clinical trial, 30 children with spastic hemiplegic CP aged six to 10years were included and randomly allocated into two groups. During six weeks, 15 patients in the intervention group received VR_TOT-based video game in addition to conventional occupational therapy, whereas 15 patients in the control group received only conventional occupational therapy. The outcomes of spoon and knife use time, as well as wrist extension range of motion (ROM), and handgrip strength were evaluated pre- and postintervention, and also at six-week follow-up. The reduction in spoon (6.27±3.08 vs 2.55±2.08; P<0.001) and knife (7.44±1.95 vs 2.74±2.46; P<0.001) usage time was significantly greater among children in the intervention group compared with the control group. The increase in grip strength was significantly higher among intervention group children (1.57±1.28 vs 0.50±0.77; P=0.011). However, the wrist extension ROM was not different between intervention and control groups. The results show that combining the VR-TOT with traditional occupational therapy methods improves fine motor function and grip strength in children with CP.

Benefits of the Global Integration Method (Método de Integração Global - MIG) in functional priorities of parents of Brazilian children and adolescents with autism spectrum disorder

BackgroundUnderstanding the priorities of parents of children and adolescents with autism spectrum disorder (ASD) is crucial for implementing evidence-based programs. This study aims to identify the functional priorities of parents of Brazilian children and adolescents with ASD, analyze variations in priorities according to the levels of support and age groups of the participants, and categorize the goals according to the categories of the International Classification of Functioning, Disability, and Health (ICF). Additionally, this study aimed to evaluate changes in parents’ performance and satisfaction with functional priorities after intervention with the Global Integration Method (Métodode Integração Global - MIG).MethodsA total of 241 children/adolescents with ASD (mean age, 6.92 ± 3.61 years) were recruited from different regions of Brazil. 76% (76%) were male, and 40.7% were classified as having support level 2. The Canadian Occupational Performance Measure was administered to parents/caregivers to identify their priorities for their children and to assess changes in performance and satisfaction with priorities after intervention with MIG. The MIG protocol consisted of functional task training in a naturalistic environment (City of Tomorrow) combined with the use of a flexible therapeutic suit (MIG Flex) and was conducted for 3 months, five times a week, for 3–4 h per day. Descriptive statistics were used to provide the priority profile. Pre- and post-intervention data were analyzed using paired t-test.ResultsParents established 1,203 functional priorities. Activities of daily living, behavioral difficulties, communication, play, and social interactions were the main functional priorities in the perception of parents/caregivers. The profiles of functional priorities were similar between the different levels of support and age groups. Approximately 64% of the priorities were classified in the activity domain of the ICF. In general, the MIG program resulted in significant improvements in performance and satisfaction for the majority of functional priorities (p < 0.05).ConclusionActivities of daily living appear to be the main priority of parents of children and adolescents with ASD, regardless of the level of support or age group. The MIG program has been associated with improvements in performance and satisfaction across several of the functional priorities identified by parents.

Use of Eye-Tracking Technology to Determine Differences Between Perceptual and Actual Navigational Performance

This study uses eye-tracking technology (ETT) to investigate discrepancies between seafarers’ perceived and actual performance during simulated maritime operations. The primary objective is to explore how misperceptions regarding the use of navigational tools—such as visual observation, radar, and ECDIS—may contribute to discrepancies in situational awareness, which is critical for safe navigation. By comparing participants’ self-reported perceptions with objective data recorded by ETT, the study highlights cognitive biases that influence navigational decision-making. Data were collected from a simulation scenario involving 32 seafarers with varying levels of maritime experience. The results reveal that participants tend to overestimate their reliance on visual observation and ECDIS, while underestimating their use of radar. These discrepancies may affect decision-making processes and could contribute to an inaccurate perception of situational awareness, although further research is needed to fully establish their direct impact on actual navigational performance. Additionally, the application of ETT identifies differences in the navigational strategies between more and less experienced seafarers, offering insights that could inform the development of training programs aimed at improving situational awareness. Statistical analyses, including Analysis of Variance (ANOVA) and Kruskal–Wallis tests, were conducted to assess the influence of demographic factors on performance. These findings suggest that ETT can be a valuable tool for identifying perceptual biases, potentially improving decision-making and enhancing training for real-world navigational tasks.

Using pretrained graph neural networks with token mixers as geometric featurizers for conformational dynamics.

Identifying informative low-dimensional features that characterize dynamics in molecular simulations remains a challenge, often requiring extensive manual tuning and system-specific knowledge. Here, we introduce geom2vec, in which pretrained graph neural networks (GNNs) are used as universal geometric featurizers. By pretraining equivariant GNNs on a large dataset of molecular conformations with a self-supervised denoising objective, we obtain transferable structural representations that are useful for learning conformational dynamics without further fine-tuning. We show how the learned GNN representations can capture interpretable relationships between structural units (tokens) by combining them with expressive token mixers. Importantly, decoupling training the GNNs from training for downstream tasks enables analysis of larger molecular graphs (that can represent small proteins at all-atom resolution) with limited computational resources. In these ways, geom2vec eliminates the need for manual feature selection and increases the robustness of simulation analyses.

Effectiveness Of Cognitive Dual Task Versus Treadmill Training On Cognition, Anxiety And Depression In Parkinson’s Patients.

Gait abnormalities can result from motor impairments caused by Parkinson's disease (PD), including functional walking. This may lead to a decrease in gait speed and stride length and an increase in stride-to-stride variability. Early motor skill development depends heavily on the cortical parts of the brain, but dual task walking may demand attention, impairing the basal ganglia circuitry that controls the more natural walking task and increasing the risk of falls. This study aims at evaluating the effectiveness of cognitive dual task on cognition, anxiety, and depression in Parkinson’s patients, to see the effectiveness of treadmill training on cognition, anxiety, and depression in Parkinson’s patients and to compare the effectiveness of cognitive dual task and treadmill training on cognition, anxiety, and depression in Parkinson’s patients.

Use of Augmented Reality for Training Assistance in Laparoscopic Surgery: Scoping Literature Review.

Laparoscopic surgery training is a demanding process requiring technical and nontechnical skills. Surgical training has evolved from traditional approaches to the use of immersive digital technologies such as virtual, augmented, and mixed reality. These technologies are now integral to laparoscopic surgery training. This scoping literature review aimed to analyze the current augmented reality (AR) solutions used in laparoscopic surgery training. Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines, we conducted a scoping review using 4 databases: Scopus, IEEE Xplore, PubMed, and ACM. Inclusion and exclusion criteria were applied to select relevant articles. Exclusion criteria were studies not using AR, not focused on laparoscopic surgery, not focused on training, written in a language other than English, or not providing relevant information on the topics studied. After selecting the articles, research questions (RQs) were formulated to guide the review. In total, 2 independent reviewers then extracted relevant data, and a descriptive analysis of the results was conducted. Of 246 initial records, 172 (69.9%) remained after removing duplicates. After applying the exclusion criteria, 76 articles were selected, with 25 (33%) later excluded for not meeting quality standards, leaving 51 (67%) in the final review. Among the devices analyzed (RQ 1), AR video-based devices were the most prevalent (43/51, 84%). The most common information provided by AR devices (RQ 1) focused on task execution and patient-related data, both appearing in 20% (10/51) of studies. Regarding sensorization (RQ 2), most studies (46/51, 90%) incorporated some form of sensorized environment, with computer vision being the most used technology (21/46, 46%) and the trainee the most frequently sensorized element (41/51, 80%). Regarding training setups (RQ 3), 39% (20/51) of the studies used commercial simulators, and 51% (26/51) made use of artificial models. Concerning the evaluation methods (RQ 4), objective evaluation was the most used, featured in 71% (36/51) of the studies. Regarding tasks (RQ 5), 43% (22/51) of studies focused on full surgical procedures, whereas 57% (29/51) focused on simple training tasks, with suturing being the most common among the latter (11/29, 38%). This scoping review highlights the evolving role of AR technologies in laparoscopic surgery training, although the impact of optical see-through devices remains unclear due to their limited use. It underscores the potential of emerging technologies such as haptic feedback, computer vision, and eye tracking to further enhance laparoscopic skill acquisition. While most relevant articles from other databases were included, some studies may have been missed due to the specific databases and search strategies used. Moreover, the need for standardized evaluation metrics is emphasized, paving the way for future research into AR's full potential in laparoscopic skill acquisition.

Human-like face pareidolia emerges in deep neural networks optimized for face and object recognition.

The human visual system possesses a remarkable ability to detect and process faces across diverse contexts, including the phenomenon of face pareidolia--seeing faces in inanimate objects. Despite extensive research, it remains unclear why the visual system employs such broadly tuned face detection capabilities. We hypothesized that face pareidolia results from the visual system's optimization for recognizing both faces and objects. To test this hypothesis, we used task-optimized deep convolutional neural networks (CNNs) and evaluated their alignment with human behavioral signatures and neural responses, measured via magnetoencephalography (MEG), related to pareidolia processing. Specifically, we trained CNNs on tasks involving combinations of face identification, face detection, object categorization, and object detection. Using representational similarity analysis, we found that CNNs that included object categorization in their training tasks represented pareidolia faces, real faces, and matched objects more similarly to neural responses than those that did not. Although these CNNs showed similar overall alignment with neural data, a closer examination of their internal representations revealed that specific training tasks had distinct effects on how pareidolia faces were represented across layers. Finally, interpretability methods revealed that only a CNN trained for both face identification and object categorization relied on face-like features-such as 'eyes'-to classify pareidolia stimuli as faces, mirroring findings in human perception. Our results suggest that human-like face pareidolia may emerge from the visual system's optimization for face identification within the context of generalized object categorization.

CADS: A Systematic Literature Review on the Challenges of Abstractive Dialogue Summarization

Abstractive dialogue summarization is the task of distilling conversations into informative and concise summaries. Although focused reviews have been conducted on this topic, there is a lack of comprehensive work that details the core challenges of dialogue summarization, unifies the differing understanding of the task, and aligns proposed techniques, datasets, and evaluation metrics with the challenges. This article summarizes the research on Transformer-based abstractive summarization for English dialogues by systematically reviewing 1262 unique research papers published between 2019 and 2024, relying on the Semantic Scholar and DBLP databases. We cover the main challenges present in dialog summarization (i.e., language, structure, comprehension, speaker, salience, and factuality) and link them to corresponding techniques such as graph-based approaches, additional training tasks, and planning strategies, which typically overly rely on BART-based encoder-decoder models. Recent advances in training methods have led to substantial improvements in language-related challenges. However, challenges such as comprehension, factuality, and salience remain difficult and present significant research opportunities. We further investigate how these approaches are typically analyzed, covering the datasets for the subdomains of dialogue (e.g., meeting, customer service, and medical), the established automatic metrics (e.g., ROUGE), and common human evaluation approaches for assigning scores and evaluating annotator agreement. We observe that only a few datasets (i.e., SAMSum, AMI, DialogSum) are widely used. Despite its limitations, the ROUGE metric is the most commonly used, while human evaluation, considered the gold standard, is frequently reported without sufficient detail on the inter-annotator agreement and annotation guidelines. Additionally, we discuss the possible implications of the recently explored large language models and conclude that our described challenge taxonomy remains relevant despite a potential shift in relevance and difficulty.

Adaptive Impedance Control of a Human–Robotic System Based on Motion Intention Estimation and Output Constraints

The rehabilitation exoskeleton represents a typical human–robot system featuring complex nonlinear dynamics. This paper is devoted to proposing an adaptive impedance control strategy for a rehabilitation exoskelton. The patient’s motion intention is estimated online by the neural network (NN) to cope with the intervention of the patient’s subjective motor awareness in the late stage of rehabilitation training. Due to the differences in impedance parameters for training tasks in individual patients and periods, the least square method was used to learn the impedance parameters of the patient. Considering the uncertainties of the exoskeleton and the safety of rehabilitation training, an adaptive neural network impedance controller with output constraints was designed. The NN was applied to approximate the unknown dynamics and the barrier Lyapunov function was applied to prevent the system from violating the output rules. The feasibility and effectiveness of the proposed strategy were verified by simulation.

Can self-distancing benefit adjustment to bereavement? A multi-method investigation.

Bereaved individuals experiencing prolonged grief often struggle to accept loss and manage emotional distress, making them a key focus of grief interventions. Self-distancing, where individuals process adverse events from an observer's perspective, may play a protective role in grief adaptation. This paper presents three studies examining self-distancing in grief adaptation. Study 1 surveyed bereaved adults (N=207) and found self-distancing was significantly and negatively correlated with grief symptoms and rumination. Study 2 used a one-session perspective manipulation exercise and showed self-distancing significantly decreased negative emotions, while self-immersion (first-person) decreased positive emotions when recalling bereavement. Study 3 implemented a brief intervention where participants with elevated grief completed an expressive writing exercise using a self-distancing perspective (n=16) or a self-immersion perspective (n=15) for 15min per day for three days. Results indicated that the two writing training tasks increased self-distancing and decreased grief symptoms, but only the self-distancing group reduced grief rumination. These findings highlight the protective role of self-distancing in grief adaptation. Low-intensity interventions, such as expressive writing from a self-distancing perspective, show promise in alleviating grief rumination and prolonged grief symptoms. These results highlight the potential of self-distancing as an accessible and effective strategy for managing prolonged grief, offering a valuable addition to existing grief interventions.

A tale of two ages: fluid reasoning as a predictor of working memory training efficacy in middle-aged and older adults

ABSTRACT Research on working memory (WM) training reveals significant variability in training effects, indicating that pretraining cognitive abilities might account for these differences. However, consensus on whether higher (magnification account) or lower (compensation account) pretraining abilities predict greater training effects remains elusive. Our study aimed to clarify the role of fluid reasoning in predicting training performance (i.e. training scores at each session) and gains on near transfer WM tasks. We conducted two studies: Study 1 focused on middle-aged adults (47–65 years) and Study 2 on older adults (65–83 years). Participants in both studies were randomly assigned to either adaptive n-back training or an active control group and have all completed three WM tasks before and after 20 training sessions – the trained n-back task and two structurally different untrained tasks. Generally, greater average training scores were found in individuals with higher fluid reasoning for both age groups, although this trend did not reach statistical significance in older adults. Similarly, higher fluid reasoning predicted greater training gains only in the sample of middle-aged adults. Further analysis showed that both, middle-aged and older participants in the training groups exhibited higher gains on the trained n-back task but not on two other WM tasks. Additionally, fluid reasoning predicted n-back gains in both the training and control group. Consistent with a growing body of research, our results show limited generalization of training effects across untrained tasks. It seems that factors beyond pretraining ability should be considered when explaining between-participant differences in training performance.

TRENDS IN THE DEVELOPMENT OF DISTANCE LEARNING TECHNOLOGIES

The article analyses the peculiarities of the introduction of distance education in Ukraine and the prospects for its development in the national educational space. The methodological features of the introduction of distance education in the context of the competence approach are considered. The object of research is the development of distance education in Ukraine. The article examines global trends in the development of the world education system, analyses the main milestones in the introduction of distance and blended education, identifies key stakeholders in distance learning and their goals, and focuses on the need to develop new didactic tools, in particular, smart technologies and smart manuals for an effective distance learning system. It is proposed to start this work with the development of the concept of a smart textbook and the design of relevant training blocks. It is explained that in the modern world this is impossible without continuous professional development, because everything around us is constantly improving and scientific and technological progress is constantly moving forward. And the same technology, on the one hand, guarantees quality training with the participation of top-level professional teachers, on the other hand, adjusts the teaching methodology depending on the level of students' training and their psychological characteristics revealed in previous tests. third, the correction of teaching methods. Learning takes place at home, on a computer, at a time convenient for the student. It has been determined that distance education is the technology of the future. It can be used in any educational system to fulfil a wide range of educational, training, upbringing and personal tasks. For this reason, new achievements in distance education need to be developed and implemented. It is even possible that in the future this technology will replace existing schools, universities and other educational institutions. The following methods were used in this study: theoretical analysis of educational and methodological literature on the organisation of distance learning and the use of smart technologies to improve the efficiency of the educational process.

Characterization of Fitness Profiles in Youth Soccer Players in Response to Playing Roles Through Principal Component Analysis

Background/Objectives: Physical fitness in youth soccer impacts individual and team performance through the specific demands that must be met on the field. Therefore, this study aimed to characterize and identify youth soccer players with regard to the roles they play on the field. Material and Methods: A cross-sectional study was designed to characterize and identify the physical fitness levels of youth soccer players using previously validated measurement tests. A total of 36 players were evaluated (15 defenders and 24 attackers) using various physical fitness tests: Squat Jump (SJ), Countermovement Jump (CMJ), Single-leg Countermovement Jumps (SLCMJs), COD-Timer 5-0-5, Speed (5, 10, 15, and 20 m), Yo-Yo Intermittent Recovery Test Level I (YYIR1), and Running-Based Anaerobic Sprint Test (RAST). The data were confirmed using the Shapiro–Wilk test. Effect sizes were obtained using the Rank-Biserial coefficient, and, to identify the profiles of attackers and defenders, principal component analysis (PCA) was employed. Results: For the strength variables, attackers obtained better results than defenders in the variable flight time in the SJ (p = 0.03; R-b = −0.33) and contact time (%) in the SLCMJ test (p = 0.04; R-b = −0.33). Meanwhile, defenders achieved better results than attackers in the SLCMJ test for the variable flight time (%) (p = 0.01; R-b = 0.33) and breaking angle (A°) in the Nordic Hamstring (p = 0.01; R-b = 0.33). The results showed differences according to the players’ roles. Three principal components were identified for both attackers and defenders. The PC1 for attackers considered variables of strength, asymmetry, change of direction, and power. PC2 only considered strength and power variables. PC3 considered variables of strength, speed, endurance, and power. For defenders, PC1 considered strength, asymmetry, and power. PC2 analyzed variables of strength, asymmetry, change of direction and power. Finally, PC3 only grouped speed variables. Conclusions: Although youth soccer positions involve offensive and defensive roles, this study reveals differences in certain physical fitness variables. Therefore, it is necessary to tailor training tasks according to the specificity of the playing position, in line with the systems of play used and the predominance of the role that players occupy, whether in defense or attack.

The Impact of Task Difficulty, Environmental Color Complexity, and Teaching Models on AR-Assisted Subway Maintenance Training

This paper investigates whether different task difficulties, color complexity, and choice of instructional application mode affect cognitive load, usability, visual discomfort, and job performance in subway overhaul training tasks and compares the findings under AR learning modes with those under traditional paper-based learning manuals. This study had sixteen participants and used eight experimental conditions. Tests were completed for environmental color complexity, task difficulty, and instructional application mode. The experimental variables were analyzed using three-way analysis of variance. The results showed that the color complexity of the carriage environment did not affect the participants’ cognitive load and performance, but the use of AR glasses in subway maintenance was influenced by task difficulty and instructional application mode. For example, the AR image format reduced task completion time by an average of 25.9% for easy tasks, while the AR real-time format reduced head deflections by an average of 83.9%, suggesting that there is reliable potential for applying AR to subway overhaul training. This paper can provide a theoretical basis for developing training strategies for the different environments and tasks difficulties of subway maintenance.

Training method and difficulty modulate electrophysiological correlates of visual perceptual learning.

The present study focused on the influence of training methods and task difficulty on event-related potentials (ERPs) at early and later visual perceptual learning (VPL) on a coherent motion identification task. Sixty participants were randomly divided into four groups for training with an adaptive stimulus (staircase group) and three constant stimuli (moderate, easy and difficult intensity groups). Visual performance improved in the staircase and moderate training groups but not in the easy or difficult training groups. ERP results revealed a decreased P1 amplitude in all groups. Additionally, staircase training increased the frontal P2 amplitude; accordingly, moderate constant stimulus training reduced the frontal P2 amplitude and increased the frontal N2 amplitude. Importantly, the change in frontal P2 amplitude was correlated with improved performance, indicating the involvement of cortices responsible for higher-order cognitive processes in VPL. Additionally, the difference in frontal P2 amplitude changes suggests the modulation of training methods (adaptive and consistent) on the role of attention in VPL. Furthermore, although behavior changes were not observed, the brains in the easy and difficult groups still presented different ERP changes. In summary, the results provide electrophysiological evidence for the modulation of training methods and task difficulty in VPL-related neuroplasticity.

VIRTUAL REALITY REHABILITATION VERSUS TASK-ORIENTED TRAINING FOR IMPROVING DYNAMIC BALANCE AND GAIT AMONG PARKINSON’S DISEASE PATIENTS

Background: Gait and balance are significant complained in patients with Parkinson’s disease (PD) which decrease their movement capability and life quality. Two innovative therapeutic modalities reported to provide better functioning in PD are virtual reality (VR) and task-oriented training (TOT). Objective : To determine effect of virtual reality rehabilitation versus task-oriented training for improving dynamic balance and gait among parkinsons disease patients Material and methods : This was comparative randomized controlled trial . Allied hospital Civil hospital Faisalabad Moeen physiotherapy clinic were utilized as study setting .Diagnosed PD patients (Hoehn and Yahr stages 2–3) aged 50–70 years were included. Exclusion criteria was patients with severe dementia or other neurological abnormalities. Sample Size was 28. Group A undergo VR-based sessions vs Group B participants were given task-oriented training Berg Balance Scale (BBS) and Timed Up and Go Test and Functional Gait Assessment (FGA) used as assessment tools. Statistical analysis performed by SPSS version 21. Results : Demonstrated BBS Score was 48.9 ± 2.9 with p score 0.03 FGA Score 23.3 ± 2.1 with p score 0.04 and 12.0 ± 1.4 TUG score with p value 0.02 among group A and group B revealed BBS Score was 46.3 ± 3.0 FGA Score was 21.7 ± 2.4 12.9 ± 1.6 was TUG score. Group A revealed more significant results as compared to group B. Conclusion: Both intervention groups revealed significant improvement after treatment but virtual reality training group showed more improvement on balance and gait among parkinsonism patients as compared to task oriented group.

Federated Learning and Reputation-Based Node Selection Scheme for Internet of Vehicles

With the rapid development of in-vehicle communication technology, the Internet of Vehicles (IoV) is gradually becoming a core component of next-generation transportation networks. However, ensuring the activity and reliability of IoV nodes remains a critical challenge. The emergence of blockchain technology offers new solutions to the problem of node selection in IoV. Nevertheless, traditional blockchain networks may suffer from malicious nodes, which pose security threats and disrupt normal blockchain operations. To address the issues of low participation and security risks among IoV nodes, this paper proposes a federated learning (FL) scheme based on blockchain and reputation value changes. This scheme encourages active involvement in blockchain consensus and facilitates the selection of trustworthy and reliable IoV nodes. First, we avoid conflicts between computing power for training and consensus by constructing state-channel transitions to move training tasks off-chain. Task rewards are then distributed to participating miner nodes based on their contributions to the FL model. Second, a reputation mechanism is designed to measure the reliability of participating nodes in FL, and a Proof of Contribution Consensus (PoCC) algorithm is proposed to allocate node incentives and package blockchain transactions. Finally, experimental results demonstrate that the proposed incentive mechanism enhances node participation in training and successfully identifies trustworthy nodes.

Speech Emotion Recognition Model Based on Joint Modeling of Discrete and Dimensional Emotion Representation

This paper introduces a novel joint model architecture for Speech Emotion Recognition (SER) that integrates both discrete and dimensional emotional representations, allowing for the simultaneous training of classification and regression tasks to improve the comprehensiveness and interpretability of emotion recognition. By employing a joint loss function that combines categorical and regression losses, the model ensures balanced optimization across tasks, with experiments exploring various weighting schemes using a tunable parameter to adjust task importance. Two adaptive weight balancing schemes, Dynamic Weighting and Joint Weighting, further enhance performance by dynamically adjusting task weights based on optimization progress and ensuring balanced emotion representation during backpropagation. The architecture employs parallel feature extraction through independent encoders, designed to capture unique features from multiple modalities, including Mel-frequency Cepstral Coefficients (MFCC), Short-term Features (STF), Mel-spectrograms, and raw audio signals. Additionally, pre-trained models such as Wav2Vec 2.0 and HuBERT are integrated to leverage their robust latent features. The inclusion of self-attention and co-attention mechanisms allows the model to capture relationships between input modalities and interdependencies among features, further improving its interpretability and integration capabilities. Experiments conducted on the IEMOCAP dataset using a leave-one-subject-out approach demonstrate the model’s effectiveness, with results showing a 1–2% accuracy improvement over classification-only models. The optimal configuration, incorporating the joint architecture, dynamic weighting, and parallel processing of multimodal features, achieves a weighted accuracy of 72.66%, an unweighted accuracy of 73.22%, and a mean Concordance Correlation Coefficient (CCC) of 0.3717. These results validate the effectiveness of the proposed joint model architecture and adaptive balancing weight schemes in improving SER performance.

Transcriptional determinants of goal-directed learning and representational drift in the parahippocampal cortex.

Task learning involves learning associations between stimuli and outcomes and storing these relationships in memory. While this information can be reliably decoded from population activity, individual neurons encoding this representation can drift over time. The circuit or molecular mechanisms underlying this drift and its role in learning are unclear. We performed two-photon calcium imaging in the perirhinal cortex during task training. Using post hoc spatial transcriptomics, we measured immediate-early gene (IEG) expression and assigned monitored neurons to excitatory or inhibitory subtypes. We discovered an IEG-defined network spanning multiple subtypes that form stimulus-outcome associations. Targeted deletion of brain-derived neurotrophic factor in the perirhinal cortex disrupted IEG expression and impaired task learning. Representational drift slowed with prolonged training. Pre-existing representations were strengthened while stimulus-reward associations failed to form. Our findings reveal the cell types and molecules regulating long-term network stability that is permissive for task learning and memory allocation.