检索结果-南通市图书馆

deep visual-linguistic fusion network considering cross-modal inconsistency for rumor detection

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2023年第12期66卷 16-32页

作者： Yang YANG Ran BAO Weili GUO de-chuan zhan Yilong YIN Jian YANG School of Computer Science and Engineering Nanjing University of Science and Technology National Key Laboratory for Novel Software Technology Nanjing University School of Software Shandong University

With the development of the Internet, users can freely publish posts on various social media platforms, which offers great convenience for keeping abreast of the world. However, posts usually carry many rumors, which require plenty of manpower for monitoring. Owing to the success of modern machine learning techniques, especially deep learning models, we tried to detect rumors as a classification problem automatically. Early attempts have always focused on building classifiers relying on image or text information, i.e., single modality in posts. Thereafter, several multimodal detection approaches employ an early or late fusion operator for aggregating multiple source information. Nevertheless, they only take advantage of multimodal embeddings for fusion and ignore another important detection factor, i.e., the intermodal inconsistency between modalities. To solve this problem, we develop a novel deep visual-linguistic fusion network(DVLFN) considering cross-modal inconsistency, which detects rumors by comprehensively considering modal aggregation and contrast information. Specifically, the DVLFN first utilizes visual and textual deep encoders, i.e., Faster R-CNN and bidirectional encoder representations from transformers, to extract global and regional embeddings for image and text modalities. Then, it predicts posts’ authenticity from two aspects:(1) intermodal inconsistency, which employs the Wasserstein distance to efficiently measure the similarity between regional embeddings of different modalities, and(2) modal aggregation, which experimentally employs the early fusion to aggregate two modal embeddings for prediction. Consequently, the DVLFN can compose the final prediction based on the modal fusion and inconsistency measure. Experiments are conducted on three real-world multimedia rumor detection datasets collected from Reddit, Good News, and Weibo. The results validate the superior performance of the proposed DVLFN.

关键词： multimodal learning Wasserstein distance rumor detection

LINDA: multi-agent local information decomposition for awareness of teammates

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2023年第8期66卷 152-168页

作者： Jiahan CAO Lei YUAN Jianhao WANG Shaowei zhanG Chongjie zhanG Yang YU de-chuan zhan National Key Laboratory for Novel Software Technology Nanjing University Institute for Interdisciplinary Information Sciences Tsinghua University Polixir Technologies

In cooperative multi-agent reinforcement learning(MARL), where agents only have access to partial observations, efficiently leveraging local information is critical. During long-time observations, agents can build awareness for teammates to alleviate the restriction of partial observability. However, previous MARL methods usually neglect awareness learning from local information for better collaboration. To address this problem, we propose a novel framework, multi-agent local information decomposition for awareness of teammates(LINDA), with which agents learn to decompose local information and build awareness for each teammate. We model the awareness as stochastic random variables and perform representation learning to ensure the informativeness of awareness representations by maximizing the mutual information between awareness and the actual trajectory of the corresponding agent. LINDA is agnostic to specific algorithms and can be flexibly integrated with different MARL methods. Sufficient experiments show that the proposed framework learns informative awareness from local partial observations for better collaboration and significantly improves the learning performance, especially on challenging tasks.

关键词： multi-agent system reinforcement learning teammates awareness centralized training with decentralized execution (CTde) StarCraft II

同方期刊数据库

PyCIL: a Python toolbox for class-incremental learning

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2023年第9期66卷 291-292页

作者： Da-Wei ZHOU Fu-Yun WANG Han-Jia YE de-chuan zhan State Key Laboratory for Novel Software Technology Nanjing University

With the rapid development of deep learning, current deep models can learn a fixed number of classes with high performance. However, in our ever-changing world, data often come from the open environment, which is with stream format or available temporarily due to privacy issues. As a result, the classification model should learn new classes incrementally instead of restarting the training process.

关键词：

TV100:a TV series dataset that pre-trained CLlP has not seen

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 Computer Science 2024年第5期18卷 237-238页

作者： Da-Wei ZHOU Zhi-Hong QI Han-Jia YE de-chuan zhan National Key Laboratory for Novel Sofware Technology Nanjing UniversityNanjing 210023China School of Artificial Intelligence Nanjing UniversityNanjing 2i0023China

In recent years,the field of deep learning has experienced remarkable growth,leading to the emergence of large,pretrained models such as ChatGPT[1],which demonstrates significant capability in understanding and responding to human language inputs,and DALL-E[2],which creatively generates images from textual descriptions in a zero-shot manner.

关键词： such textual see

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

基于模型分解的小样本学习

中国科学：信息科学 2020年第5期50卷 662-674页

作者：叶翰嘉詹德川计算机软件新技术国家重点实验室(南京大学) 南京210023

机器学习依赖大量样本的统计信息进行模型的训练,从而能对未知样本进行精准的预测.搜集样本及标记需要耗费大量的资源,因而如何基于少量样本(few-shot learning)进行模型的训练至关重要.有效的模型先验(prior)能够降低模型训练对样本的... 详细信息

机器学习依赖大量样本的统计信息进行模型的训练,从而能对未知样本进行精准的预测.搜集样本及标记需要耗费大量的资源,因而如何基于少量样本(few-shot learning)进行模型的训练至关重要.有效的模型先验(prior)能够降低模型训练对样本的需求.本文基于元学习(meta learning)框架,从相关的、类别不同的数据中学习模型先验,并将这种先验应用于新类别的少样本任务.与此同时,本文提出"模型组合先验"(MCP,model composition prior)方法,通过目标函数的最优条件对模型结构进行分解,并分别估计模型的各个组成部分,得到有效的分类器.这种分解方式具有较高的可解释性,能够指导在不同小样本任务中"共享"与"独立"的成分,从而指导元学习的具体实现.在人造数据中,本文方法能够恢复出小样本任务之间的关联性;在图像数据上,MCP方法能取得比当前主流方法更优异的效果.

关键词：小样本学习元学习模型先验模型分解

维普期刊数据库

Reducing energy barriers of chemical reactions with a nanomicrocell catalyst consisting of integrated active sites in conductive matrices

在线全文

学校读者我要写书评

暂无评论

Science Bulletin 2019年第6期64卷 385-390页

作者： Guo-Peng zhan chuan-de State Key Laboratory of Silicon Materials Center for Chemistry of High-Performance&Novel MaterialsDepartment of ChemistryZhejiang University

Reducing energy barriers of chemical reactions is the never-ending endeavor of *** by the high reactivity of primary cells,we develop a nanosized fuel cell catalyst(denoted as nanomicrocell catalyst),consisting of integrated electrode pairs,conductive matrices and electrolytes,to improve the chemical ***,the anodes are Pd species which is combining with the electron-rich N atoms in B-and-N co-doped carbon dots;the cathodes are electron-deficient B atoms;and the conductive matrices are B-and-N co-doped carbon dots which are connecting with the electrode *** to the reactivity of primary cells,the catalytic properties of the nanomicrocell catalyst in hydrogenation of benzaldehyde are depending on the properties of electrode pairs,conductive matrices and *** unique catalytic properties are attributed to the different substrate adsorption capability and catalytic properties of paired electrodes,and the easy migration of electrons and charge carriers,which could improve the synergetic effect between paired active ***,this work may open up a new window for designed synthesis of advanced catalysts which could highly lower the energy barriers of chemical reactions.

关键词： Nanomicrocell catalysts Conductive matrices Electrode pairs Electrolytes Primary cells

维普期刊数据库

Revolutionizing breast cancer detection:Advances in fluorescent probe technology

在线全文

学校读者我要写书评

暂无评论

Nano Research 2025年第1期18卷 653-674页

作者： Qian Gao Lin Tang Jun zhan Yiyuan Xue Feng Luo State Key Laboratory of Oral Diseases&National Center for Stomatology&National Clinical Research Center for Oral Diseases West China Hospital of StomatologySichuan UniversityChengdu 610041China Emergency Obstetrics and Gynecology West China Second University HospitalSichuan UniversityChengdu 610041China department of Obstetrics and Gynecology West China Second University HospitalSichuan UniversityChengdu 610041China department of Prosthodontics West China School of StomatologySichuan UniversityChengdu 610041China

With the increasing incidence of breast cancer globally,there is an urgent demand for enhanced diagnostic and therapeutic *** comprehensive review explores advancements in fluorescent probe technology for breast cancer imaging and early *** probes offer unparalleled sensitivity and specificity in detecting cancer biomarkers,enabling early diagnosis and delineation of tumor margins during *** categorize these probes into molecular,nano,ratiometric,the second near-infrared window(NIR-II),and multifunctional varieties,each with unique capabilities for targeting specific cancer *** review highlights recent innovations in probe design,emphasizing their applications in pathogenesis studies,drug development,precision medicine,and fluorescence-guided *** their promising clinical potential,challenges such as safety,biomarker targeting accuracy,system compatibility,and clinical translation *** these hurdles is crucial for the integration of fluorescent probes into standard cancer care,aiming to improve treatment outcomes and patient quality of life.

关键词： breast cancer fluorescent probes imaging early diagnosis biomarker

Generation of global 1-km daily top-of-atmosphere outgoing longwave radiation product from 2000 to 2021 using machine learning

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

International Journal of Digital Earth 2023年第1期16卷 2002-2012页

作者： chuan zhan Shunlin Liang School of Resources and Environmental Engineering Wuhan University of Science and TechnologyWuhanPeople's Republic of China department of Geography The University of Hong KongPokfulamPeople's Republic of China

Top-of-atmosphere(TOA)outgoing longwave radiation(OLR),a key component of the Earth’s energy budget,serves as a diagnostic of the Earth’s climate system response to incoming solar ***,existing products are typically estimated using broadband sensors with coarse spatial *** paper presents a machine learning method to estimate TOA OLR by directly linking Moderate Resolution Imaging Spectroradiometer(MODIS)TOA radiances with TOA OLR determined by Clouds and the Earth’s Radiant Energy System(CERES)and other information,such as the viewing geometry,land surface temperature and cloud top temperature determined by Modern-Era Retrospective analysis for Research and Applications,Version 2(MERRA-2).Models are built separately under clear-and cloudy-sky conditions using a gradient boosting regression *** test results show that the root mean square errors(RMSEs)of the clear-sky and cloudy-sky models for estimating instantaneous values are 4.1 and 7.8 W/m^(2),***-time conversion ratios derived from CERES daily and hourly OLR data are used to convert the instantaneous MODIS OLR to daily ***-comparisons of the daily results show that the RMSE of the estimated MODIS OLR is 8.9 W/m^(2) in East *** developed high resolution dataset will be beneficial in analyzing the regional energy budget.

关键词： TOA outgoing longwave radiation MODIS CERES machine learning Earth’s energy budget

MECHANISM deSIGN AND MOTION ANALYSIS OF A SPHERICAL MOBILE ROBOT

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

Chinese Journal of Mechanical Engineering 2005年第4期18卷 542-545页

作者： zhan Qiang Jia chuan Ma Xiaohui Zhai Yutao Robot Research Institute Beihang University Beijing 100083 China China Aerospace Science ＆ Industry Academy of Technology Beijing 100037 China

A new spherical mobile robot BHQ-1 is designed. The spherical robot is driven by two internally mounted motors that induce the ball to move straight and turn around on a fiat surface. A dynamic model of the robot is developed with Lagrange method and factors affecting the driving torque of two motors are analyzed. The relationship between the turning radius of the robot and the length of two links is discussed in order to optimize its mechanism design. Simulation and experimental results demonstrate the good controllability and motion performance of BHQ-1.

关键词： Spherical mobile robot Mechanism design Motion analysis

维普期刊数据库