Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Re...Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.展开更多
The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in S...The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in Spanish is challenging due to linguistic complexity and the scarcity of annotated resources.In this paper,we compare two predominant AI-based approaches for the forensic detection of malicious hate speech:(1)finetuning encoder-only models that have been trained in Spanish and(2)In-Context Learning techniques(Zero-and Few-Shot Learning)with large-scale language models.Our approach goes beyond binary classification,proposing a comprehensive,multidimensional evaluation that labels each text by:(1)type of speech,(2)recipient,(3)level of intensity(ordinal)and(4)targeted group(multi-label).Performance is evaluated using an annotated Spanish corpus,standard metrics such as precision,recall and F1-score and stability-oriented metrics to evaluate the stability of the transition from zero-shot to few-shot prompting(Zero-to-Few Shot Retention and Zero-to-Few Shot Gain)are applied.The results indicate that fine-tuned encoder-only models(notably MarIA and BETO variants)consistently deliver the strongest and most reliable performance:in our experiments their macro F1-scores lie roughly in the range of approximately 46%–66%depending on the task.Zero-shot approaches are much less stable and typically yield substantially lower performance(observed F1-scores range approximately 0%–39%),often producing invalid outputs in practice.Few-shot prompting(e.g.,Qwen 38B,Mistral 7B)generally improves stability and recall relative to pure zero-shot,bringing F1-scores into a moderate range of approximately 20%–51%but still falling short of fully fine-tuned models.These findings highlight the importance of supervised adaptation and discuss the potential of both paradigms as components in AI-powered cybersecurity and malware forensics systems designed to identify and mitigate coordinated online hate campaigns.展开更多
This study examines how foreign language education in the artificial intelligence(AI)era could assist the cultivation of national consciousness through a technology-enhanced pedagogy of film appreciation.Using The Wil...This study examines how foreign language education in the artificial intelligence(AI)era could assist the cultivation of national consciousness through a technology-enhanced pedagogy of film appreciation.Using The Wild Robot as a case study,we argue that cinematic narratives serve as cultural mirrors,offering immersive,reflective,and affective sites for intercultural learning.We propose a three-layered pedagogical framework-progressing from semiotic decoding,through narrative and value comparison,to creative identity construction-that integrates intelligent tools to develop both communicative competence and an agentive sense of belonging.The approach exemplifies a humanistic turn in language teaching,aiming to form“rooted global communicators”who can engage in cross-civilization dialogue with cultural confidence and critical awareness.展开更多
<p align="left"> <span style="font-family:Verdana;">Online learning has been on an upward trend for many years and is becoming more and more prevalent every day, consistently presenting...<p align="left"> <span style="font-family:Verdana;">Online learning has been on an upward trend for many years and is becoming more and more prevalent every day, consistently presenting the less privileged parts of our society with an equal opportunity at education. Unfortunately, though, it seldom takes advantage of the new technologies and capabilities offered by the modern World Wide Web. In this article, we present an interactive online platform that provides users with learning activities for students of English as a foreign language. The platform focuses on using audiovisual multimedia content and a user experience (UX) centered approach to provide learners with an enhanced learning experience that aims at improving their knowledge level while at the same time increasing their engagement and motivation to participate in learning. To achieve this, the platform uses advanced techniques, such as interactive vocabulary and pronunciation assistance, mini-games, embedded media, voice recording, and more. In addition, the platform provides educators with analytics about user engagement and performance. In this study, more than 100 young students participated in a preliminary use of the aforementioned platform and provided feedback concerning their experience. Both the platform’s metrics and the user-provided feedback indicated increased engagement and a preference of the participants for interactive audiovisual multimedia-based learning activities.</span> </p>展开更多
Fundamental physics often confronts complex symbolic problems with few guiding exemplars or established principles.While artificial intelligence(AI)offers promise,its typical need for vast datasets to learn from hinde...Fundamental physics often confronts complex symbolic problems with few guiding exemplars or established principles.While artificial intelligence(AI)offers promise,its typical need for vast datasets to learn from hinders its use in these information-scarce frontiers.We introduce learning at criticality(LaC),a reinforcement learning scheme that tunes large language models(LLMs)to a sharp learning transition,addressing this information scarcity.At this transition,LLMs achieve peak generalization from minimal data,exemplified by 7-digit base-7 addition-a test of nontrivial arithmetic reasoning.To elucidate this peak,we analyze a minimal concept-network model designed to capture the essence of how LLMs might link tokens.Trained on a single exemplar,this model also undergoes a sharp learning transition.This transition exhibits hallmarks of a second-order phase transition,notably power-law distributed solution path lengths.At this critical point,the system maximizes a“critical thinking pattern”crucial for generalization,enabled by the underlying scale-free exploration.This suggests LLMs reach peak performance by operating at criticality,where such explorative dynamics enable the extraction of underlying operational rules.We demonstrate LaC in quantum field theory:an 8B-parameter LLM,tuned to its critical point by LaC using a few exemplars of symbolic Matsubara sums,solves unseen,higher-order problems,significantly outperforming far larger models.LaC thus leverages critical phenomena,a physical principle,to empower AI for complex,data-sparse challenges in fundamental physics.展开更多
The emergence of embodied cognition theory has altered our traditional understanding of children’s language learning,emphasizing the close connection between the body,environment,and movement.This paper discusses the...The emergence of embodied cognition theory has altered our traditional understanding of children’s language learning,emphasizing the close connection between the body,environment,and movement.This paper discusses the opportunities,challenges,and future directions of research on children’s language learning from the perspective of embodied cognition.It concludes that multisensory engagement can greatly improve children’s comprehension and memorization of language knowledge and that language acquisition is intimately tied to bodily perception,movement,and emotional experience.In addition,children’s language acquisition can also be effectively aided by embodied cognition techniques as multimedia aids,gesture and enactment,and imagery.Based on previous evidence,we propose an integrated language learning framework and a new relevance-integration taxonomy for children’s language learning from the perspectives of embodied cognition and cognitive load theories.In order to support the long-term growth of children’s language education,future research should focus more on the requirement of embodied language learning in the preschool-primary transition and optimize the teaching objectives and contents.展开更多
The paper aims to examine the application of multimedia technology in expanding vocabulary in second language acquisition.Incorporating innovative technology such as mobile applications,gaming applications,websites,an...The paper aims to examine the application of multimedia technology in expanding vocabulary in second language acquisition.Incorporating innovative technology such as mobile applications,gaming applications,websites,and other related online tools has increased learners’vocabulary mastery,engagement,and motivation levels.Interactional processes like media-embedded objects,teach-learning capacity algorithms,and feedback help learners receive the course in a personalized way that considers individual learning patterns or abilities.However,there are the following challenges:accessibility issues,total reliance on technology,and issues related to privacy.The following challenges affecting learning that arise from using gadgets:the digital divide,limited device access,and environmental issues that may distract a learner in a technology-enabled environment.Moreover,the security issue for data and the ethical question of users’information remain important too.Hence,the paper provides arguments that although these technologies contribute significantly to vocabulary acquisition,the challenge that emerges should be addressed by integrating technology in teaching and learning alongside conventional methods for vocabulary acquisition,which is a practical language acquisition tool that should not be monopolized.展开更多
Under the tide of economic globalization,college English teaching should not only focus on the improvement of language ability,but also on the cultivation of students’critical thinking ability.This paper takes the in...Under the tide of economic globalization,college English teaching should not only focus on the improvement of language ability,but also on the cultivation of students’critical thinking ability.This paper takes the integration of language learning and critical thinking ability as the breakthrough point,explores the college English teaching mode under the background of the integration of the two,analyzes the current situation and disadvantages of the separation of the two in the current teaching,and puts forward the integration path from the aspects of curriculum design,teacher training,evaluation system,and so on.With the help of activities such as creating real language situations,carrying out debates and critical reading,it helps students strengthen the improvement of logical analysis and critical thinking ability in their gradual learning,realize the coordinated development of language learning and critical thinking ability,and cultivate compound talents with both language literacy and critical thinking ability for the society.展开更多
Federated Learning enables privacy-preserving training of Transformer-based language models,but remains vulnerable to backdoor attacks that compromise model reliability.This paper presents a comparative analysis of de...Federated Learning enables privacy-preserving training of Transformer-based language models,but remains vulnerable to backdoor attacks that compromise model reliability.This paper presents a comparative analysis of defense strategies against both classical and advanced backdoor attacks,evaluated across autoencoding and autoregressive models.Unlike prior studies,this work provides the first systematic comparison of perturbation-based,screening-based,and hybrid defenses in Transformer-based FL environments.Our results show that screening-based defenses consistently outperform perturbation-based ones,effectively neutralizing most attacks across architectures.However,this robustness comes with significant computational overhead,revealing a clear trade-off between security and efficiency.By explicitly identifying this trade-off,our study advances the understanding of defense strategies in federated learning and highlights the need for lightweight yet effective screening methods for trustworthy deployment in diverse application domains.展开更多
Thermophilic proteins maintain their structure and function at high temperatures,making them widely useful in industrial applications.Due to the complexity of experimental measurements,predicting the melting temperatu...Thermophilic proteins maintain their structure and function at high temperatures,making them widely useful in industrial applications.Due to the complexity of experimental measurements,predicting the melting temperature(T_(m))of proteins has become a research hotspot.Previous methods rely on amino acid composition,physicochemical properties of proteins,and the optimal growth temperature(OGT)of hosts for T_(m)prediction.However,their performance in predicting T_(m)values for thermophilic proteins(T_(m)>60℃)are generally unsatisfactory due to data scarcity.Herein,we introduce T_(m)Pred,a T_(m)prediction model for thermophilic proteins,that combines protein language model,graph convolutional network and Graphormer module.For performance evaluation,T_(m)Pred achieves a root mean square error(RMSE)of 5.48℃,a pearson correlation coefficient(P)of 0.784,and a coefficient of determination(R~2)of 0.613,representing improvements of 19%,15%,and 32%,respectively,compared to the state-of-the-art predictive models like DeepTM.Furthermore,T_(m)Pred demonstrated strong generalization capability on independent blind test datasets.Overall,T_(m)Pred provides an effective tool for the mining and modification of thermophilic proteins by leveraging deep learning.展开更多
This paper investigates the capabilities of large language models(LLMs)to leverage,learn and create knowledge in solving computational fluid dynamics(CFD)problems through three categories of baseline problems.These ca...This paper investigates the capabilities of large language models(LLMs)to leverage,learn and create knowledge in solving computational fluid dynamics(CFD)problems through three categories of baseline problems.These categories include(1)conventional CFD problems that can be solved using existing numerical methods in LLMs,such as lid-driven cavity flow and the Sod shock tube problem;(2)problems that require new numerical methods beyond those available in LLMs,such as the recently developed Chien-physics-informed neural networks for singularly perturbed convection-diffusion equations;and(3)problems that cannot be solved using existing numerical methods in LLMs,such as the ill-conditioned Hilbert linear algebraic systems.The evaluations indicate that reasoning LLMs overall outperform non-reasoning models in four test cases.Reasoning LLMs show excellent performance for CFD problems according to the tailored prompts,but their current capability in autonomous knowledge exploration and creation needs to be enhanced.展开更多
Drug repurposing offers a promising alternative to traditional drug development and significantly re-duces costs and timelines by identifying new therapeutic uses for existing drugs.However,the current approaches ofte...Drug repurposing offers a promising alternative to traditional drug development and significantly re-duces costs and timelines by identifying new therapeutic uses for existing drugs.However,the current approaches often rely on limited data sources and simplistic hypotheses,which restrict their ability to capture the multi-faceted nature of biological systems.This study introduces adaptive multi-view learning(AMVL),a novel methodology that integrates chemical-induced transcriptional profiles(CTPs),knowledge graph(KG)embeddings,and large language model(LLM)representations,to enhance drug repurposing predictions.AMVL incorporates an innovative similarity matrix expansion strategy and leverages multi-view learning(MVL),matrix factorization,and ensemble optimization techniques to integrate heterogeneous multi-source data.Comprehensive evaluations on benchmark datasets(Fdata-set,Cdataset,and Ydataset)and the large-scale iDrug dataset demonstrate that AMVL outperforms state-of-the-art(SOTA)methods,achieving superior accuracy in predicting drug-disease associations across multiple metrics.Literature-based validation further confirmed the model's predictive capabilities,with seven out of the top ten predictions corroborated by post-2011 evidence.To promote transparency and reproducibility,all data and codes used in this study were open-sourced,providing resources for pro-cessing CTPs,KG,and LLM-based similarity calculations,along with the complete AMVL algorithm and benchmarking procedures.By unifying diverse data modalities,AMVL offers a robust and scalable so-lution for accelerating drug discovery,fostering advancements in translational medicine and integrating multi-omics data.We aim to inspire further innovations in multi-source data integration and support the development of more precise and efficient strategies for advancing drug discovery and translational medicine.展开更多
This review interrogates empirical and theoretical research on agentic engagement in foreign language(FL)learning.Through synthesizing peer-reviewed studies from Web of Science and CNKI databases,it maps the theoretic...This review interrogates empirical and theoretical research on agentic engagement in foreign language(FL)learning.Through synthesizing peer-reviewed studies from Web of Science and CNKI databases,it maps the theoretical evolution,methodological innovations,key influencing factors and proposed suggestion for further research on student agency.Future research should prioritize,longitudinal studies,culturally comparative designs,validity constructs and ethical evaluations of artificial intelligence’s impact on learner autonomy.This review calls for a holistic approach to FL education,where agentic engagement bridges individual initiative,pedagogical innovation,and sociocultural responsiveness to empower learners in multilingual global contexts.展开更多
Accessible communication based on sign language recognition(SLR)is the key to emergency medical assistance for the hearing-impaired community.Balancing the capture of both local and global information in SLR for emerg...Accessible communication based on sign language recognition(SLR)is the key to emergency medical assistance for the hearing-impaired community.Balancing the capture of both local and global information in SLR for emergency medicine poses a significant challenge.To address this,we propose a novel approach based on the inter-learning of visual features between global and local information.Specifically,our method enhances the perception capabilities of the visual feature extractor by strategically leveraging the strengths of convolutional neural network(CNN),which are adept at capturing local features,and visual transformers which perform well at perceiving global features.Furthermore,to mitigate the issue of overfitting caused by the limited availability of sign language data for emergency medical applications,we introduce an enhanced short temporal module for data augmentation through additional subsequences.Experimental results on three publicly available sign language datasets demonstrate the efficacy of the proposed approach.展开更多
Native language has been rejected for a long time in the foreign language class, which results from a misunderstanding to native language transfer by most teachers. Actually, it is an effective learning strategy to co...Native language has been rejected for a long time in the foreign language class, which results from a misunderstanding to native language transfer by most teachers. Actually, it is an effective learning strategy to complete the communication task under the help of native language, which should be acknowledged. Using native language to explain specific words or grammar rules can take advantage of the limited class-time efficiently and increase the class efficiency; using native language to discuss teaching method and solve the problems for the students can promote their enthusiasm. In these specific teaching processes, native language is an important teaching resource. Instances identify that native language has positive effects in the foreign language learning and teaching and it should have its own standpoint.展开更多
As the ultimate goal of education, autonomy in language learning has aroused a lot of attention from scholars at home and abroad. While in universities of China, students do not have strong autonomy in English languag...As the ultimate goal of education, autonomy in language learning has aroused a lot of attention from scholars at home and abroad. While in universities of China, students do not have strong autonomy in English language learning. The author tries to adopt specific meta-cognitive strategies to facilitate students＇ autonomy in learning by improving learners＇ capacities in study planning or management, monitoring and evaluating in learning to raise their consciousness and ability in autonomy, and lay a foundation for life-long learning.展开更多
The paper, with the backdrop of web-based autonomous learning put forward by the recent college English teaching reform, aims to explore teachers＇ roles in this learning process in students＇ perception through the m...The paper, with the backdrop of web-based autonomous learning put forward by the recent college English teaching reform, aims to explore teachers＇ roles in this learning process in students＇ perception through the means of questionnaires and interviews. It further analyzes the possible reasons why students perceive their teachers＇ roles in such a way, in the hope of providing some implications for web-based college English autonomous learning.展开更多
The thesis introduces a comparative study of students'autonomous listening practice in a web-based autonomous learning center and the traditional teacher-dominated listening practice in a traditional language lab....The thesis introduces a comparative study of students'autonomous listening practice in a web-based autonomous learning center and the traditional teacher-dominated listening practice in a traditional language lab.The purpose of the study is to find how students'listening strategies differ in these two approaches and thereby to find which one better facilitates students'listening proficiency.展开更多
文摘Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.
基金the research project LaTe4PoliticES(PID2022-138099OB-I00)funded by MCIN/AEI/10.13039/501100011033 and the European Fund for Regional Development(ERDF)-a way to make Europe.Tomás Bernal-Beltrán is supported by University of Murcia through the predoctoral programme.
文摘The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in Spanish is challenging due to linguistic complexity and the scarcity of annotated resources.In this paper,we compare two predominant AI-based approaches for the forensic detection of malicious hate speech:(1)finetuning encoder-only models that have been trained in Spanish and(2)In-Context Learning techniques(Zero-and Few-Shot Learning)with large-scale language models.Our approach goes beyond binary classification,proposing a comprehensive,multidimensional evaluation that labels each text by:(1)type of speech,(2)recipient,(3)level of intensity(ordinal)and(4)targeted group(multi-label).Performance is evaluated using an annotated Spanish corpus,standard metrics such as precision,recall and F1-score and stability-oriented metrics to evaluate the stability of the transition from zero-shot to few-shot prompting(Zero-to-Few Shot Retention and Zero-to-Few Shot Gain)are applied.The results indicate that fine-tuned encoder-only models(notably MarIA and BETO variants)consistently deliver the strongest and most reliable performance:in our experiments their macro F1-scores lie roughly in the range of approximately 46%–66%depending on the task.Zero-shot approaches are much less stable and typically yield substantially lower performance(observed F1-scores range approximately 0%–39%),often producing invalid outputs in practice.Few-shot prompting(e.g.,Qwen 38B,Mistral 7B)generally improves stability and recall relative to pure zero-shot,bringing F1-scores into a moderate range of approximately 20%–51%but still falling short of fully fine-tuned models.These findings highlight the importance of supervised adaptation and discuss the potential of both paradigms as components in AI-powered cybersecurity and malware forensics systems designed to identify and mitigate coordinated online hate campaigns.
基金supported by the project:Hunan Provincial Educational Science Research Project“Research on Cultivating National Consciousness in College Foreign Language Courses(XJT23CGD001)”.
文摘This study examines how foreign language education in the artificial intelligence(AI)era could assist the cultivation of national consciousness through a technology-enhanced pedagogy of film appreciation.Using The Wild Robot as a case study,we argue that cinematic narratives serve as cultural mirrors,offering immersive,reflective,and affective sites for intercultural learning.We propose a three-layered pedagogical framework-progressing from semiotic decoding,through narrative and value comparison,to creative identity construction-that integrates intelligent tools to develop both communicative competence and an agentive sense of belonging.The approach exemplifies a humanistic turn in language teaching,aiming to form“rooted global communicators”who can engage in cross-civilization dialogue with cultural confidence and critical awareness.
文摘<p align="left"> <span style="font-family:Verdana;">Online learning has been on an upward trend for many years and is becoming more and more prevalent every day, consistently presenting the less privileged parts of our society with an equal opportunity at education. Unfortunately, though, it seldom takes advantage of the new technologies and capabilities offered by the modern World Wide Web. In this article, we present an interactive online platform that provides users with learning activities for students of English as a foreign language. The platform focuses on using audiovisual multimedia content and a user experience (UX) centered approach to provide learners with an enhanced learning experience that aims at improving their knowledge level while at the same time increasing their engagement and motivation to participate in learning. To achieve this, the platform uses advanced techniques, such as interactive vocabulary and pronunciation assistance, mini-games, embedded media, voice recording, and more. In addition, the platform provides educators with analytics about user engagement and performance. In this study, more than 100 young students participated in a preliminary use of the aforementioned platform and provided feedback concerning their experience. Both the platform’s metrics and the user-provided feedback indicated increased engagement and a preference of the participants for interactive audiovisual multimedia-based learning activities.</span> </p>
基金supported by the National Key Research and Development Program of China(Grant No.2024YFA1408604 for K.C.and X.C.)the National Natural Science Foundation of China(Grant Nos.12047503,12447103 for K.C.and X.C.,12325501 for P.Z.,and 12275263 for Y.D.and S.H.)+1 种基金the Innovation Program for Quantum Science and Technology(Grant No.2021ZD0301900 for Y.D.and S.H.)the Natural Science Foundation of Fujian Province of China(Grant No.2023J02032 for Y.D.and S.H.)。
文摘Fundamental physics often confronts complex symbolic problems with few guiding exemplars or established principles.While artificial intelligence(AI)offers promise,its typical need for vast datasets to learn from hinders its use in these information-scarce frontiers.We introduce learning at criticality(LaC),a reinforcement learning scheme that tunes large language models(LLMs)to a sharp learning transition,addressing this information scarcity.At this transition,LLMs achieve peak generalization from minimal data,exemplified by 7-digit base-7 addition-a test of nontrivial arithmetic reasoning.To elucidate this peak,we analyze a minimal concept-network model designed to capture the essence of how LLMs might link tokens.Trained on a single exemplar,this model also undergoes a sharp learning transition.This transition exhibits hallmarks of a second-order phase transition,notably power-law distributed solution path lengths.At this critical point,the system maximizes a“critical thinking pattern”crucial for generalization,enabled by the underlying scale-free exploration.This suggests LLMs reach peak performance by operating at criticality,where such explorative dynamics enable the extraction of underlying operational rules.We demonstrate LaC in quantum field theory:an 8B-parameter LLM,tuned to its critical point by LaC using a few exemplars of symbolic Matsubara sums,solves unseen,higher-order problems,significantly outperforming far larger models.LaC thus leverages critical phenomena,a physical principle,to empower AI for complex,data-sparse challenges in fundamental physics.
基金supported by the Humanities and Social Science Fund of the Ministry of Education of China(23YJA190012)Guangdong Provincial College Student Innovation and Entrepreneurship Training Program(S202410577105)+1 种基金Huizhou Philosophy and Social Sciences Discipline Co-Construction Project(HZ2023GJ128)Characteristic Innovation Project of Colleges and Universities in Guangdong Province(2021WTSCX090).
文摘The emergence of embodied cognition theory has altered our traditional understanding of children’s language learning,emphasizing the close connection between the body,environment,and movement.This paper discusses the opportunities,challenges,and future directions of research on children’s language learning from the perspective of embodied cognition.It concludes that multisensory engagement can greatly improve children’s comprehension and memorization of language knowledge and that language acquisition is intimately tied to bodily perception,movement,and emotional experience.In addition,children’s language acquisition can also be effectively aided by embodied cognition techniques as multimedia aids,gesture and enactment,and imagery.Based on previous evidence,we propose an integrated language learning framework and a new relevance-integration taxonomy for children’s language learning from the perspectives of embodied cognition and cognitive load theories.In order to support the long-term growth of children’s language education,future research should focus more on the requirement of embodied language learning in the preschool-primary transition and optimize the teaching objectives and contents.
基金Interim Achievements of the“Yingying Technology Empowerment–Application-Oriented Talent Enhancement Project at Changchun College of Electronic Technology”under the Fourth Phase of the 2024 Ministry of Education’s Employment-Education Collaboration Project(Project Number:2024121188944Project Leader:Chunhua Ren)+3 种基金Interim Achievements of the“Directional Cultivation Project for Composite Talents at Changchun College of Electronic Technology”under the Fourth Phase of the 2024 Ministry of Education’s Supply-Demand Matching and Employment-Education Cultivation Program(Project Number:2024121107571Project Leader:Chunhua Ren)Interim Achievements of the“Research on the Cultivation Path of Craftsmanship Spirit among University Teachers in the Context of Industry-University Collaboration”under the 2025 Ministry of Education’s Industry-University Cooperative Education Project(Project Number:2505164755Project Leader:Chunhua Ren)。
文摘The paper aims to examine the application of multimedia technology in expanding vocabulary in second language acquisition.Incorporating innovative technology such as mobile applications,gaming applications,websites,and other related online tools has increased learners’vocabulary mastery,engagement,and motivation levels.Interactional processes like media-embedded objects,teach-learning capacity algorithms,and feedback help learners receive the course in a personalized way that considers individual learning patterns or abilities.However,there are the following challenges:accessibility issues,total reliance on technology,and issues related to privacy.The following challenges affecting learning that arise from using gadgets:the digital divide,limited device access,and environmental issues that may distract a learner in a technology-enabled environment.Moreover,the security issue for data and the ethical question of users’information remain important too.Hence,the paper provides arguments that although these technologies contribute significantly to vocabulary acquisition,the challenge that emerges should be addressed by integrating technology in teaching and learning alongside conventional methods for vocabulary acquisition,which is a practical language acquisition tool that should not be monopolized.
文摘Under the tide of economic globalization,college English teaching should not only focus on the improvement of language ability,but also on the cultivation of students’critical thinking ability.This paper takes the integration of language learning and critical thinking ability as the breakthrough point,explores the college English teaching mode under the background of the integration of the two,analyzes the current situation and disadvantages of the separation of the two in the current teaching,and puts forward the integration path from the aspects of curriculum design,teacher training,evaluation system,and so on.With the help of activities such as creating real language situations,carrying out debates and critical reading,it helps students strengthen the improvement of logical analysis and critical thinking ability in their gradual learning,realize the coordinated development of language learning and critical thinking ability,and cultivate compound talents with both language literacy and critical thinking ability for the society.
基金supported by a research fund from Chosun University,2024.
文摘Federated Learning enables privacy-preserving training of Transformer-based language models,but remains vulnerable to backdoor attacks that compromise model reliability.This paper presents a comparative analysis of defense strategies against both classical and advanced backdoor attacks,evaluated across autoencoding and autoregressive models.Unlike prior studies,this work provides the first systematic comparison of perturbation-based,screening-based,and hybrid defenses in Transformer-based FL environments.Our results show that screening-based defenses consistently outperform perturbation-based ones,effectively neutralizing most attacks across architectures.However,this robustness comes with significant computational overhead,revealing a clear trade-off between security and efficiency.By explicitly identifying this trade-off,our study advances the understanding of defense strategies in federated learning and highlights the need for lightweight yet effective screening methods for trustworthy deployment in diverse application domains.
基金financially supported by the National Key R&D Program of China(Nos.2020YFA0908100 and 2023YFF1204401)Shenzhen Medical Research Fund(No.B2302037)+1 种基金the National Natural Science Foundation of China(Nos.22331003 and 21925102)Beijing National Laboratory for Molecular Sciences(No.BNLMS-CXXM-202006)。
文摘Thermophilic proteins maintain their structure and function at high temperatures,making them widely useful in industrial applications.Due to the complexity of experimental measurements,predicting the melting temperature(T_(m))of proteins has become a research hotspot.Previous methods rely on amino acid composition,physicochemical properties of proteins,and the optimal growth temperature(OGT)of hosts for T_(m)prediction.However,their performance in predicting T_(m)values for thermophilic proteins(T_(m)>60℃)are generally unsatisfactory due to data scarcity.Herein,we introduce T_(m)Pred,a T_(m)prediction model for thermophilic proteins,that combines protein language model,graph convolutional network and Graphormer module.For performance evaluation,T_(m)Pred achieves a root mean square error(RMSE)of 5.48℃,a pearson correlation coefficient(P)of 0.784,and a coefficient of determination(R~2)of 0.613,representing improvements of 19%,15%,and 32%,respectively,compared to the state-of-the-art predictive models like DeepTM.Furthermore,T_(m)Pred demonstrated strong generalization capability on independent blind test datasets.Overall,T_(m)Pred provides an effective tool for the mining and modification of thermophilic proteins by leveraging deep learning.
基金supported by the National Natural Science Foundation of China Basic Science Center Program for“Multiscale Problems in Nonlinear Mechanics”(Grant No.11988102)the National Natural Science Foundation of China(Grant No.12202451).
文摘This paper investigates the capabilities of large language models(LLMs)to leverage,learn and create knowledge in solving computational fluid dynamics(CFD)problems through three categories of baseline problems.These categories include(1)conventional CFD problems that can be solved using existing numerical methods in LLMs,such as lid-driven cavity flow and the Sod shock tube problem;(2)problems that require new numerical methods beyond those available in LLMs,such as the recently developed Chien-physics-informed neural networks for singularly perturbed convection-diffusion equations;and(3)problems that cannot be solved using existing numerical methods in LLMs,such as the ill-conditioned Hilbert linear algebraic systems.The evaluations indicate that reasoning LLMs overall outperform non-reasoning models in four test cases.Reasoning LLMs show excellent performance for CFD problems according to the tailored prompts,but their current capability in autonomous knowledge exploration and creation needs to be enhanced.
基金supported by the National Natural Science Foundation of China(Grant No.:62101087)the China Postdoctoral Science Foundation(Grant No.:2021MD703942)+2 种基金the Chongqing Postdoctoral Research Project Special Funding,China(Grant No.:2021XM2016)the Science Foundation of Chongqing Municipal Commission of Education,China(Grant No.:KJQN202100642)the Chongqing Natural Science Foundation,China(Grant No.:cstc2021jcyj-msxmX0834).
文摘Drug repurposing offers a promising alternative to traditional drug development and significantly re-duces costs and timelines by identifying new therapeutic uses for existing drugs.However,the current approaches often rely on limited data sources and simplistic hypotheses,which restrict their ability to capture the multi-faceted nature of biological systems.This study introduces adaptive multi-view learning(AMVL),a novel methodology that integrates chemical-induced transcriptional profiles(CTPs),knowledge graph(KG)embeddings,and large language model(LLM)representations,to enhance drug repurposing predictions.AMVL incorporates an innovative similarity matrix expansion strategy and leverages multi-view learning(MVL),matrix factorization,and ensemble optimization techniques to integrate heterogeneous multi-source data.Comprehensive evaluations on benchmark datasets(Fdata-set,Cdataset,and Ydataset)and the large-scale iDrug dataset demonstrate that AMVL outperforms state-of-the-art(SOTA)methods,achieving superior accuracy in predicting drug-disease associations across multiple metrics.Literature-based validation further confirmed the model's predictive capabilities,with seven out of the top ten predictions corroborated by post-2011 evidence.To promote transparency and reproducibility,all data and codes used in this study were open-sourced,providing resources for pro-cessing CTPs,KG,and LLM-based similarity calculations,along with the complete AMVL algorithm and benchmarking procedures.By unifying diverse data modalities,AMVL offers a robust and scalable so-lution for accelerating drug discovery,fostering advancements in translational medicine and integrating multi-omics data.We aim to inspire further innovations in multi-source data integration and support the development of more precise and efficient strategies for advancing drug discovery and translational medicine.
文摘This review interrogates empirical and theoretical research on agentic engagement in foreign language(FL)learning.Through synthesizing peer-reviewed studies from Web of Science and CNKI databases,it maps the theoretical evolution,methodological innovations,key influencing factors and proposed suggestion for further research on student agency.Future research should prioritize,longitudinal studies,culturally comparative designs,validity constructs and ethical evaluations of artificial intelligence’s impact on learner autonomy.This review calls for a holistic approach to FL education,where agentic engagement bridges individual initiative,pedagogical innovation,and sociocultural responsiveness to empower learners in multilingual global contexts.
基金supported by the National Natural Science Foundation of China(No.62376197)the Tianjin Science and Technology Program(No.23JCYBJC00360)the Tianjin Health Research Project(No.TJWJ2025MS045).
文摘Accessible communication based on sign language recognition(SLR)is the key to emergency medical assistance for the hearing-impaired community.Balancing the capture of both local and global information in SLR for emergency medicine poses a significant challenge.To address this,we propose a novel approach based on the inter-learning of visual features between global and local information.Specifically,our method enhances the perception capabilities of the visual feature extractor by strategically leveraging the strengths of convolutional neural network(CNN),which are adept at capturing local features,and visual transformers which perform well at perceiving global features.Furthermore,to mitigate the issue of overfitting caused by the limited availability of sign language data for emergency medical applications,we introduce an enhanced short temporal module for data augmentation through additional subsequences.Experimental results on three publicly available sign language datasets demonstrate the efficacy of the proposed approach.
文摘Native language has been rejected for a long time in the foreign language class, which results from a misunderstanding to native language transfer by most teachers. Actually, it is an effective learning strategy to complete the communication task under the help of native language, which should be acknowledged. Using native language to explain specific words or grammar rules can take advantage of the limited class-time efficiently and increase the class efficiency; using native language to discuss teaching method and solve the problems for the students can promote their enthusiasm. In these specific teaching processes, native language is an important teaching resource. Instances identify that native language has positive effects in the foreign language learning and teaching and it should have its own standpoint.
文摘As the ultimate goal of education, autonomy in language learning has aroused a lot of attention from scholars at home and abroad. While in universities of China, students do not have strong autonomy in English language learning. The author tries to adopt specific meta-cognitive strategies to facilitate students＇ autonomy in learning by improving learners＇ capacities in study planning or management, monitoring and evaluating in learning to raise their consciousness and ability in autonomy, and lay a foundation for life-long learning.
文摘The paper, with the backdrop of web-based autonomous learning put forward by the recent college English teaching reform, aims to explore teachers＇ roles in this learning process in students＇ perception through the means of questionnaires and interviews. It further analyzes the possible reasons why students perceive their teachers＇ roles in such a way, in the hope of providing some implications for web-based college English autonomous learning.
文摘The thesis introduces a comparative study of students'autonomous listening practice in a web-based autonomous learning center and the traditional teacher-dominated listening practice in a traditional language lab.The purpose of the study is to find how students'listening strategies differ in these two approaches and thereby to find which one better facilitates students'listening proficiency.
