检索结果-南通市图书馆

Model gradient: unified model and policy learning in model-based reinforcement learning

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2024年第4期18卷 117-128页

作者： Chengxing JIA Fuxiang zhang Tian XU jing-Cheng PANG Zongzhang zhang Yang YU National Key Laboratory for Novel Software Technology Nanjing UniversityNanjing 210023China Polixir Technologies Nanjing 210000China

Model-based reinforcement learning is a promising direction to improve the sample efficiency of reinforcement learning with learning a model of the *** model learning methods aim at fitting the transition data,and commonly employ a supervised learning approach to minimize the distance between the predicted state and the real *** supervised model learning methods,however,diverge from the ultimate goal of model learning,i.e.,optimizing the learned-in-the-model *** this work,we investigate how model learning and policy learning can share the same objective of maximizing the expected return in the real *** find model learning towards this objective can result in a target of enhancing the similarity between the gradient on generated data and the gradient on the real *** thus derive the gradient of the model from this target and propose the Model Gradient algorithm(MG)to integrate this novel model learning approach with policy-gradient-based policy *** conduct experiments on multiple locomotion control tasks and find that MG can not only achieve high sample efficiency but also lead to better convergence performance compared to traditional model-based reinforcement learning approaches.

关键词： reinforcement learning model-based reinforcement learning Markov decision process

A comparison of statistical learning of naturalistic textures between DCNNs and the human visual hierarchy

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Technological Sciences) 2024年第8期67卷 2310-2318页

作者： LU XinCheng YUAN ZiQi zhang YiChi AI HaiLin CHENG SiYuan GE YiRan fang fang CHEN NiHong Department of Psychological and Cognitive Sciences Tsinghua UniversityBeijing 100084China IDG/McGovern Institute for Brain Research Tsinghua UniversityBeijing 100084China Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China School of Psychological and Cognitive Sciences and Beijing Key Laboratory of Behavior and Mental Health Peking UniversityBeijing 100871China Key Laboratory of Machine Perception(Ministry of Education) Peking UniversityBeijing 100871China Peking-Tsinghua Center for Life Sciences Peking UniversityBeijing 100871China IDG/McGovern Institute for Brain Research Peking UniversityBeijing 100871China State Key Laboratory of Brain and Cognitive Science Institute of PsychologyChinese Academy of SciencesBeijing 100101China

The visual system continuously adapts to the statistical properties of the environment. Existing evidence shows a close resemblance between deep convolutional neural networks(CNNs) and primate visual stream in neural selectivity to naturalistic textures above the primary visual processing stage. This study delves into the mechanisms of perceptual learning in CNNs,focusing on how they assimilate the high-order statistics of natural textures. Our results show that a CNN model achieves a similar performance improvement as humans, as manifested in the learning pattern across different types of high-order image statistics. While L2 was the first stage exhibiting texture selectivity, we found that stages beyond L2 were critically involved in learning. The significant contribution of L4 to learning was manifested both in the modulations of texture-selective responses and in the consequences of training with frozen connection weights. Our findings highlight learning-dependent plasticity in the mid-to-high-level areas of the visual hierarchy. This research introduces an AI-inspired approach for studying learning-induced cortical plasticity, utilizing DCNNs as an experimental framework to formulate testable predictions for empirical brain studies.

关键词： CNN perceptual learning naturalistic texture psychophysics

Relative exchangeable copper， a high-quality biomarker for differentiation of Traditional Chinese Medicine syndrome in Wilson’s disease

在线全文

学校读者我要写书评

暂无评论

Journal of Traditional Chinese Medicine 2024年第5期44卷 1017-1023页

作者： DING Yufeng YANG Wenming CHENG Yuan YANG Yulong ZHU Jun LU Yachun fang Xiang zhang jing Department of Neurology， The First Affiliated Hospital of Anhui University of Traditional Chinese Medicine Department of Clinical Laboratory， the First Affiliated Hospital of Anhui University of Traditional Chinese Medicine

Objective: To investigate the value of relative exchangeable copper (REC) in diagnosing Wilson’s disease (WD) and to determine its significance in the differentiation of Traditional Chinese Medicine (TCM) syndrome. Methods: A total of 78 patients with WD were recruited on the same day of the medical visit， and among them， 32 were suffering from non-WD (N-WD) and 37 were heterozygous ATP7B carriers (HC) enrolled as controls. Molecular genetic testing was performed for diagnosing WD and HC. Additionally， REC levels in different TCM syndromes were investigated. The correlation between REC and serum ceruloplasmin (Cp)， serum copper oxidase (CO)， and 24-h urinary copper was analyzed using the Global Assessment Scale and Unified Wilson’s Disease Rating Scale and the significance of REC in WD diagnosis was investigated. Resting-state functional magnetic resonance imaging was used to assess the clinical symptoms of WD and analyze its severity in different TCM syndromes. Results: REC determination helped in significantly distinguishing patients with N-WD or HC from those with WD with a cut-off of 21.15%. Furthermore， the comparative analysis of REC ratios among different TCM syndromes showed markedly high REC levels in the dampness-heat internal accumulation syndrome group. Additionally， the seed-based functional connectivity value from the putamen to the cerebellum was signi?cantly correlated with TCM syndromes. Conclusion: REC level is a high-quality biomarker for diagnosing WD that effectively reflects disease severity and plays an essential role in deciding on treatment strategies and prognosis. Furthermore， REC levels are closely related to TCM syndromes in WD; thus， it is a potential objective quantitative indicator for distinguishing TCM syndromes in WD.

关键词： Global Assessment Scale Traditional Chinese Medicine syndrome Wilson’s disease relative exchangeable copper the Unified Wilson’s Disease Rating Scale

Efficacy and Safety of Sanjie Analgesic Capsule in Patients with Endometriosis-Associated Pain:A Multicenter,3:1 Randomized,Double-Blind,Placebo-Controlled Trial

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Integrative Medicine 2024年第9期30卷 780-787页

作者： LENG Jin-hua DUAN Hua GUAN Zheng ZHOU Ying-fang QU Hong XU Kai-hong zhang Shao-fen zhang Qin WANG Xin LIN Kai-qing LANG jing-he Department of Obstetrics and Gynecology Peking Union Medical College HospitalBeijing100730China Department of Gynecology Beijing Obstetrics and Gynecology HospitalCapital Medical UniversityBeijing100006China Department of Gynecology Chinese PLA General HospitalBeijing100039China Department of Gynecology Peking University First HospitalBeijing100034China Department of Gynecology Beijing Chao-yang HospitalBeijing100020China Department of Gynecology Womens Hospital School of Medicine Zhejiang UniversityHangzhouZhejiang Province310006China Department of Gynecology Obstetrics and Gynecology Hospital of Fudan UniversityShanghai200090China Department of Gynecology Hangzhou Hospital of Traditional Chinese MedicineHangzhou310007China Department of Gynecology Affiliated Hospital of Liaoning University of Traditional Chinese MedicineShenyang110032China

Objective:To assess the efficacy and safety of Sanjie Analgesic Capsule(SAC)in Chinese patients with endometriosis-associated ***:This was a multicenter,randomized,double-blind,placebo-controlled trial conducted at 15 centers between November 2013 and July 2017 in *** 323 patients with endometriosis were randomized at a 3:1 ratio to the SAC group(241 cases)and placebo group(82 cases)by stratified block *** in the SAC or placebo groups were given SAC or placebo 1.6 g 3 times per day,orally,respectively since the first day of menstruation for 3 consecutive menstrual *** primary endpoint was clinical response to dysmenorrhea evaluated using a 10-point Visual Analogue Scale at 3 and 6 *** secondary endpoint was the pain score evaluated by VAS(chronic pelvic pain,defecation pain,and dyspareunia)at 3 and 6 months,and the pain recurrence rate at 6 *** events(AEs)were recorded during the ***:A total of 241 women were included in the SAC group,and 82 were in the placebo *** these women,217(90.0%)and 71(86.6%)completed the intervention,*** 3 months,overall response rate(ORR)was significantly higher in women administered SAC(80.1%)compared with those who received a placebo(30.5%,P<0.01).Six months after treatment,the ORR for dysmenorrhea was 62.7%in the SAC group and 31.7%in the placebo group(P<0.01).Chronic pelvic pain and defecation pain were significantly improved by SAC compared with placebo(both P<0.05).The incidence rates of total AEs events in the SAC and placebo groups were 6.6%and 9.8%,respectively,and no significant difference was shown between the two groups(P=0.339).Conclusion:SAC is well-tolerated and may improve dysmenorrhea in women with endometriosis-associated pain.

关键词： endometriosis pain Sanjie Analgesic Capsule randomized controlled trial Chinese medicine

Microphase Separation of Semiflexible Ring Diblock Copolymers

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Polymer Science 2024年第2期42卷 267-276,I0011页

作者： Dan-Yan Qin Sheng-Da Zhao Zhi-Xin Liu jing zhang Xing-Hua zhang School of Physical Science and Engineering Beijing Jiaotong UniversityBeijing100044China

Aiming at the difficult problem of solving the conformation statistics of complex polymers, this study presents a novel and concise conformation statistics theoretical approach based on Monte Carlo and Neural Network method. This method offers a new research idea for investigating the conformation statistics of complex polymers, characterized by its simplicity and practicality. It can be applied to more complex topological structure, more higher degree of freedom polymer systems with higher dimensions, theory research on dynamic self-consistent field theory and polymer field theory, as well as the analysis of scattering experimental data. The conformation statistics of complex polymers determine the structure and response properties of the system. Using the new method proposed in this study, taking the semiflexible ring diblock copolymer as an example, Monte Carlo simulation is used to sample this ring conformation to construct the dataset of polymer. The structure factor describing conformation statistics are expressed as continuous functions of structure parameters by neural network supervised learning. This is the innovation of this work. As an application, the structure factors represented by neural networks were introduced into the random phase approximation theory to study the microphase separation of semiflexible ring diblock copolymers. The influence of the ring's topological properties on the phase transition behavior was pointed out.

关键词： Ring copolymer Diblock copolymers Semiflexible The order-disorder transition

Maternal serum branched-chain amino acids in early pregnancy and offspring growth patterns from 1 year to 8 years of age

在线全文

学校读者我要写书评

暂无评论

中华医学杂志(英文版) 2024年第15期137卷 1876-1878页

作者： Li Ninghua Leng Junhong zhang Rui Wang Hui Li Weiqin zhang Shuang Qiao Yijuan Li jing Yu Zhijie Hu Gang fang Zhongze Yang Xilin Department of Epidemiology and Biostatistics School of Public Health Tianjin Medical University Tianjin China Project Office Tianjin Women and Children’s Health Center Tianjin China Tianjin Key Laboratory of Environment Nutrition and Public Health Tianjin China Tianjin Center for International Collaborative Research on Environment Nutrition and Public Health Tianjin China Population Cancer Research Program and Department of Pediatrics Dalhousie University Halifax Canada Chronic Disease Epidemiology Laboratory Pennington Biomedical Research Center Baton Rouge LA USA Department of Toxicology and Sanitary Chemistry School of Public Health Tianjin Medical University Tianjin China

Communication efficiency optimization of federated learning for computing and network convergence of 6G networks

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Information Technology & Electronic Engineering 2024年第5期25卷 713-727页

作者： Yizhuo CAI Bo LEI Qianying ZHAO jing PENG Min WEI Yushun zhang Xing zhang Wireless Signal Processing and Network Laboratory Beijing University of Posts and TelecommunicationsBeijing 100876China Research Institute of China Telecom Co. Ltd.Beijing 102209China Beijing Telecom Beijing 100011China

Federated learning effectively addresses issues such as data privacy by collaborating across participating devices to train global ***,factors such as network topology and computing power of devices can affect its training or communication process in complex network *** and network convergence(CNC)of sixth-generation(6G)networks,a new network architecture and paradigm with computing-measurable,perceptible,distributable,dispatchable,and manageable capabilities,can effectively support federated learning training and improve its communication *** guiding the participating devices'training in federated learning based on business requirements,resource load,network conditions,and computing power of devices,CNC can reach this *** this paper,to improve the communication eficiency of federated learning in complex networks,we study the communication eficiency optimization methods of federated learning for CNC of 6G networks that give decisions on the training process for different network conditions and computing power of participating *** simulations address two architectures that exist for devices in federated learning and arrange devices to participate in training based on arithmetic power while achieving optimization of communication efficiency in the process of transferring model *** results show that the methods we proposed can cope well with complex network situations,effectively balance the delay distribution of participating devices for local training,improve the communication eficiency during the transfer of model parameters,and improve the resource utilization in the network.

关键词： Computing and network convergence Communication efficiency Federated learning Two architectures

Targeting synaptic pathology in Alzheimer’s disease

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Zoological Research 2024年第4期45卷 875-876页

作者： jing Cao Yun-Wu zhang Fujian Provincial Key Laboratory of Neurodegenerative Disease and Aging Research Institute of NeuroscienceSchool of MedicineXiamen UniversityXiamenFujian 361102China

Alzheimer’s disease(AD)is a neurodegenerative disorder characterized by memory and cognitive *** two primary pathological hallmarks of AD include the accumulation ofβ-amyloid(Aβ)plaques and formation of neurofibrillary tangles(NFTs)composed predominantly of hyperphosphorylated tau protein(zhang et al.,2023).Moreover,synaptic dysfunction is considered another crucial factor in AD pathogenesis and is closely associated with cognitive ***,synaptic damage,loss,and dysfunction manifest earlier than the pathological features of Aβplaques and NFTs(Knopman et al.,2021).Therefore,understanding the mechanisms underlying synaptic dysfunction in AD is vital for providing insights into disease mechanisms and developing novel therapeutic strategies.

关键词： Alzheimer Aβ tau

Unlock the full potential of carbon cloth-based scaffolds towards magnesium metal storage via regulation on magnesiophilicity and surface geometric structure

在线全文

学校读者我要写书评

暂无评论

Journal of Energy Chemistry 2024年第3期90卷 423-434,I0010页

作者： jing Liu Min Wang Zhonghua zhang Jinlei zhang Yitao He Zhenfang Zhou Guicun Li College of Materials Science and Engineering Qingdao University of Science and TechnologyQingdao 266042ShandongChina Department of Pharmacy Jining Medical UniversityRizhao 276826ShandongChina Department of Energy and Power Engineering School of Energy and EnvironmentAnhui University of TechnologyMa’anshan 243000AnhuiChina

The development of rechargeable magnesium(Mg) batteries is of practical significance to upgrade the electric energy storage devices due to exceptional capacity and abundant resources of Mg-metal ***,the reversible Mg electrochemistry suffers from unsatisfied rate capability and lifespan,mainly caused by non-uniform distribution of *** this work,a fresh design concept of threedimensional carbon cloths scaffolds is proposed to overcome the uncontrollable Mg growth via homogenizing electric field and improving magnesiophilicity.A microscopic smooth and nitrogen-containing defective carbonaceous layer is constructed through a facile pyrolysis of ZIF8 on carbon *** revealed by finite element simulation and DFT calculation results,the smooth surface endows with uniform electric field distribution and simultaneously the nitrogen-doping species enable good magnesiophilicity of *** fine and uniform Mg nucleus as well as the inner electrodeposit behavior are also *** a result,an exceptional cycle life of 500 cycles at 4.0 mA cm^(-2) and 4.0 mA h cm^(-2) is firstly realized to our best ***,the functional scaffolds can be cycled for over 2200 h at 2.0 mA cm^(-2) under a normalized capacity of 5.0 mA h cm^(-2),far exceeding previous *** work offers an effective approach to enable the full potential of carbon cloths-based scaffolds towards metal storage for next generation battery applications.

关键词： Magnesium metal anodes Electrodeposition Heterogeneous nucleation