检索结果-南通市图书馆

Two-Stage Approach for Targeted Knowledge Transfer in Self-Knowledge Distillation

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2024年第11期11卷 2270-2283页

作者： Zimo Yin Jian Pu Yijie Zhou xiangyang xue the School of Computer Science Fudan University the Institute of Science and Technology for Brain Inspired Intelligence Fudan University IEEE

Knowledge distillation(KD) enhances student network generalization by transferring dark knowledge from a complex teacher network. To optimize computational expenditure and memory utilization, self-knowledge distillation(SKD) extracts dark knowledge from the model itself rather than an external teacher network. However, previous SKD methods performed distillation indiscriminately on full datasets, overlooking the analysis of representative samples. In this work, we present a novel two-stage approach to providing targeted knowledge on specific samples, named two-stage approach self-knowledge distillation(TOAST). We first soften the hard targets using class medoids generated based on logit vectors per class. Then, we iteratively distill the under-trained data with past predictions of half the batch size. The two-stage knowledge is linearly combined, efficiently enhancing model performance. Extensive experiments conducted on five backbone architectures show our method is model-agnostic and achieves the best generalization ***, TOAST is strongly compatible with existing augmentation-based regularization methods. Our method also obtains a speedup of up to 2.95x compared with a recent state-of-the-art method.

关键词： Cluster-based regularization iterative prediction refinement model-agnostic framework self-knowledge distillation(SKD) two-stage knowledge transfer

维普期刊数据库博看期刊

Metaverse:Perspectives from graphics,interactions and visualization

在线全文

学校读者我要写书评

暂无评论

Visual Informatics 2022年第1期6卷 56-67页

作者： Yuheng Zhao Jinjing Jiang Yi Chen Richen Liu Yalong Yang xiangyang xue Siming Chen School of Data Science Fudan UniversityChina Beijing Key Laboratory of Big Data Technology for Food Safety Beijing Technology and Business UniversityChina School of Computer and Electronic Information Nanjing Normal UniversityChina Department of Computer Science Virginia TechUSA

The metaverse is a visual world that blends the physical world and digital *** present,the development of the metaverse is still in the early stage,and there lacks a framework for the visual construction and exploration of the *** this paper,we propose a framework that summarizes how graphics,interaction,and visualization techniques support the visual construction of the metaverse and user-centric *** introduce three kinds of visual elements that compose the metaverse and the two graphical construction methods in a *** propose a taxonomy of interaction technologies based on interaction tasks,user actions,feedback and various sensory channels,and a taxonomy of visualization techniques that assist user *** potential applications and future opportunities are discussed in the context of visual construction and exploration of the *** hope this paper can provide a stepping stone for further research in the area of graphics,interaction and visualization in the metaverse.

关键词： Metaverse Virtual reality/Augmented reality Computer graphics User interaction Immersive visualization

Characterization of Episomal Replication of Bovine Papillomavirus Type 1 DNA in Long-Term Virion-Infected Saccharomyces Cerevisiae Culture

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Virologica Sinica 2021年第6期36卷 1492-1502页

作者： Quanmei Tu Weixu Feng Zhuo Chen Qijia Li Yu Zhao Jun Chen Pengfei Jiang xiangyang xue Lifang Zhang Kong-Nan Zhao Department of Obstetrics and Gynaecology The Second Affiliated Hospital and Yuyin Children Hospital of Wenzhou Medical UniversityWenzhou325035China Institute of Molecular Virology and Immunology Department of Microbiology and ImmunologySchool of Basic Medical SciencesWenzhou Medical UniversityWenzhou325035China Australian Institute for Bioengineering and Nanotechnology the University of QueenslandSt Lucia4067Australia

We have previously reported that bovine papillomavirus type 1(BPV-1) DNA can replicate its genome and produce infectious virus-like particles in short term virion-infected S. cerevisiae(budding yeast) cultures(Zhao and Frazer 2002,Journal of Virology, 76:3359–64 and 76:12265–73). Here, we report the episomal replications of BPV-1 DNA in long term virion-infected S. cerevisiae culture up to 108 days. Episomal replications of the BPV-1 DNA could be divided into three patterns at three stages, early active replication(day 3–16), middle weak replication(day 23–34/45) and late stable replication(day 45–82). Two-dimensional gel electrophoresis analysis and Southern blot hybridization have revealed further that multiple replication intermediates of BPV-1 DNA including linear form, stranded DNA, monomers and higher oligomers were detected in the virion-infected yeast cells over the time course. Higher oligomers shown as covalently closed circular DNAs(cccDNAs) are the most important replication intermediates that serve as the main nuclear transcription template for producing all viral RNAs in the viral life cycle. In this study, the cccDNAs were generated at the early active replication stage with the highest frequencies and then at late stable replication, but they appeared to be suppressed at the middle weak replication. Our data provided a novel insight that BPV-1 genomic DNA could replicate episomally for the long period and produce the key replication intermediates cccDNAs in S. cerevisiae system.

关键词： Bovine papillomavirus type 1(BPV-1) Saccharomyces cerevisiae Genomic DNA replication Agarose gel electrophoresis Southern blot hybridization

Genetic evidence for facial variation being a composite phenotype of cranial variation and facial soft tissue thickness

在线全文

学校读者我要写书评

暂无评论

Journal of Genetics and Genomics 2022年第10期49卷 934-942页

作者： Wei Qian Manfei Zhang Kaiwen Wan Yunxia Xie Siyuan Du Jiarui Li Xiongzheng Mu Jiange Qiu xiangyang xue Xiahai Zhuang Yingzhi Wu Fan Liu Sijia Wang State Key Laboratory of Genetic Engineering Human Phenome InstituteZhangjiang Fudan International Innovation CenterFudan UniversityShanghai 201203China School of Computer Science Fudan UniversityShanghai 200438China CAS Key Laboratory of Computational Biology Shanghai Institute of Nutrition and HealthUniversity of Chinese Academy of SciencesChinese Academy of SciencesShanghai 200031China School of Data Science Fudan UniversityShanghai 200433China Academy of Medical Science Zhengzhou UniversityZhengzhouHenan 450052China Department of Plastic Surgery Huashan HospitalFudan UniversityShanghai 200040China CAS Key Laboratory of Genomic and Precision Medicine Beijing Institute of GenomicsUniversity of Chinese Academy of SciencesBeijing 100101China School of Future Technology University of Chinese Academy of SciencesBeijing 101408China Center for Excellence in Animal Evolution and Genetics Chinese Academy of SciencesKunmingYunnan 650223China

Facial and cranial variation represent a multidimensional set of highly correlated and heritable *** is known about the genetic basis explaining this *** develop a software package ALo SFL for simultaneous localization of facial and cranial landmarks from head computed tomography(CT)images,apply it in the analysis of head CT images of 777 Han Chinese women,and obtain a set of phenotypes representing variation in face,skull and facial soft tissue thickness(FSTT).Association analysis of 301 single nucleotide polymorphisms(SNPs)from 191 distinct genomic loci previously associated with facial variation reveals an unexpected larger number of loci showing significant associations(P<1e-3)with cranial phenotypes than expected under the null(O/E=3.39),suggesting facial and cranial phenotypes share a substantial proportion of genetic *** FSTT to a SNP-only model shows a large impact in explaining facial variance.A gene ontology analysis reveals that bone morphogenesis and osteoblast differentiation likely underlie our cranial-significant ***,this study simultaneously investigates the genetic effects on both facial and cranial variation of the same sample,supporting that facial variation is a composite phenotype of cranial variation and FSTT.

关键词： Facial variation Cranial variation Association analysis Single nucleotide polymorphism Facial soft tissue thickness Computed tomography

Chimerically fused antigen rich of overlapped epitopes from latent membrane protein 2 （LMP2） of Epstein-Barr virus as a potential vaccine and diagnostic agent

在线全文

学校读者我要写书评

暂无评论

Cellular & Molecular Immunology 2016年第4期13卷 492-501页

作者： Xiaoyun Lin Shao Chen xiangyang xue Lijun Lu Shanli Zhu Wenshu Li Xiangmin Chen Xiaozhi Zhong Pengfei Jiang Torsoo Sophia Sename Yi Zheng Lifang Zhang Department of Microbiology and Immunology Institute of Molecular Virology and Immunology Wenzhou Medical University Wenzhou China School of Laboratory Medicine and Life Science Wenzhou Medical University Wenzhou China

Epstein-Barr virus （EBV） is prevalent throughout the world and is associated with several malignant diseases in humans. Latent membrane protein 2 （LMP2） of EBV plays a crucial role in the pathogenesis of EBV-associated tumors; therefore, LMP2 has been considered to be a potential immunodiagnostic and immunotherapeutic target. A multi-epitope-based antigen is a promising option for therapeutic vaccines and diagnoses of such malignancies. In this study, we systematically screened cytotoxic T lymphocyte （CTL）, helper T cell （Th） and B-cell epitopes within EBV-LMP2 using bioinformatics. Based on the screen, two peptides rich in overlapping epitopes of both T cells and B cells were selected to construct a plasmid containing the sequence for a chimeric multi-epitope protein referred to as EBV-LMP2m, which is composed of LMP2aa195-232 and LMP2aa419-436. The EBV-LMP2m protein was expressed in E. coil BL21 （DE3） after prokaryotic codon optimization. Inoculation of the purified chimeric antigen in BALB/c mice induced not only high levels of specific IgG in the serum and secretory IgA in the vaginal mucus but also a specific CTL response. By using purified EBV-LMP2m as an antigen, the presence of specific IgG in the serum specimens of 202 nasopharyngeal carcinoma （NPC） patients was effectively detected with 52.84% sensitivity and 95.40% specificity, which represents an improvement over the traditional detection method based on VCA-IgA （60.53% sensitivity and 76.86% specificity）. The above results indicate that EBV-LMP2m may be used not only as a potential target antigen for EBV-associated tumors but also a diagnostic agent for NPC patients.

关键词： Epstein-Barr virus (EBV) epitope latent membrane protein 2 (LMP2) vaccine

On challenges of AI to cognitive security and safety

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Security and Safety 2023年第2期2卷 44-56页

作者： Ruiyang Huang Xiaoqing Zheng Yuting Shang xiangyang xue National Digital Switching System Engineering&Technological R&D Center Zhengzhou 450002China Fudan University Shanghai 200433China

Recent advances in deep learning have led to disruptive breakthroughs in artificial intelligence(AI),fueling the jump in ChatGPT-like large language models(LLMs).As with any emerging technology,it is a two-sided coin,bringing not only vast social impacts but also significant security concerns,especially in the socio-cognitive *** this back-ground,this work starts with an inherent mechanism analysis of cognitive domain games,from which it proceeds to explore the security concerns facing the cognitive domain as well as to analyze the formation mechanisms of a cognitive immune ***,inspired by behavioral mimicry in biology,this work will elaborate on new approaches to cognitive security from three aspects:Mimicry Computing,Mimicry Defense,and Mimicry Intelligence.

关键词： Cognitive security Artificial intelligence Endogenous safety and security Mimicry intelligence

一种TCP友好的Internet DTV组播发送速率控制机制

在线全文

学校读者我要写书评

暂无评论

计算机研究与发展 2007年第4期44卷 623-629页

作者：李飞王新薛向阳复旦大学计算机科学与工程系上海200433

当前的Internet没有对流媒体应用提供足够的QoS的保证,同时Internet DTV的组播发送速率必须满足以下两点要求:1)能够自适应网络拥塞的变化;2)能够适应节目码率的要求,因此需要在服务器端进行组播发送速率控制.分析了Internet DTV组播的... 详细信息

当前的Internet没有对流媒体应用提供足够的QoS的保证,同时Internet DTV的组播发送速率必须满足以下两点要求:1)能够自适应网络拥塞的变化;2)能够适应节目码率的要求,因此需要在服务器端进行组播发送速率控制.分析了Internet DTV组播的难点,提出了一种基于缓冲区管理的具有网络自适应特性的组播发送速率控制方法.通过合理地控制发送方的发送速率,既能自适应网络状况的变化,又能满足流媒体实时播放的需求.仿真实验结果表明,该算法在保证服务器端发送缓冲区不溢出前提下,与单纯采用TFMCC方法比较,减少了接收端的平均丢包率,提高了网络电视节目流的传输质量.

关键词：网络数字电视组播 TFMCC 发送速率控制

在线全文

学校读者我要写书评

暂无评论

基于索引结构的高效运动视频检索

计算机研究与发展 2006年第11期43卷 1953-1958页

作者：张静路红薛向阳复旦大学计算机科学与工程系上海200433

视频片断检索是视频领域的研究热点,为了提高查询效率,利用高维索引结构Vector-Approxi-mationFile(VA-File)来组织视频子片段,并采用新的相似度模型和基于限定性滑动窗口的高效视频检索算法进行视频片段检索.提出的子片段的分割算法能... 详细信息

视频片断检索是视频领域的研究热点,为了提高查询效率,利用高维索引结构Vector-Approxi-mationFile(VA-File)来组织视频子片段,并采用新的相似度模型和基于限定性滑动窗口的高效视频检索算法进行视频片段检索.提出的子片段的分割算法能够较好地区分运动的细节动作,且相似度模型充分考虑了对应子片段之间的视觉相似性以及时间顺序关系,因此对于运动视频的检索十分有效.实验证明,对于运动视频片段检索不仅具有较高的查询效率,而且能够得到较高的查全率和准确率.

关键词：视频片段检索 VA-File 高维索引结构 k近邻查询相似度度量

MOTION COMPENSATION BASED ON MOTION VECTOR INTERPOLATION

在线全文

学校读者我要写书评

暂无评论

Journal of Electronics(China) 1998年第1期15卷 22-28页

作者： xue xiangyang Wu Lide(Department of Computer Science, Fudan University, Shanghai 200433) Department of Computer Science Fudan University Shanghai

A modified block matching algorithm (BMA) with motion correlation constraint (MCCBMA) is proposed at first. Then a novel motion compensation algorithm (INTPMC) which computes motion vector for each pixel by interpolating motion vectors is presented. In order to increase interframe prediction performance and decrease the computational complexity, an optimizing algorithm for partial motion vectors is described at last. Experimental results show that the proposed algorithm can improve the prediction performance obviously with a moderately increased complexity compared with the conventional full search block matching algorithm (FSBMA).

关键词： Motion analysis Motion compensation Video coding