作者: Yusuf, Abdulganiyu Abdu;Feng Chong;Mao Xianling (1School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China;2South-East Information Technology Institute of Beijing Institute of Technology, Beijing, China;3Beijing Engineering Research Centre of High Volume Language Information Processing and Cloud Computing Application, Beijing, China;4National Biotechnology Development Agency, Abuja, Nigeria)
出处: Artificial Intelligence Review: An International Science and Engineering Journal 2022 Vol.55 No.8 P6277-6300
关键词: Computer vision;NLP;VQA;GCN;Datasets
摘要: Graph neural network is a deep learning approach widely applied on structural and non-structural scenarios due to its substantial performance and inte ...
作者:
Fuzhi Zhang1,2;
出处: Expert Systems with Applications 2022 Vol.203 P117482
摘要: The collusive spamming behavior on e-commerce websites seriously affects the purchase decisions of consumers and disrupts the fair competition order a ...
作者:
Jinchang Ren1,2,3;
出处: Pattern Recognition Letters 2022 Vol.155 P165-170
摘要: Accurate extraction of semantic objects such as ventricles and myocardium from magnetic resonance (MR) images is one essential but very challenging ta ...
作者: Zhan, Jiaao1;Chen, Qian2;Chen, Boxing2;Wang, Wen2;Bai, Yu1;Gao, Yang1; (1School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China;2Machine Intelligence Technology Lab, Alibaba DAMO Academy, China)
出处: arXiv 2022
摘要: Non-autoregressive translation (NAT) models suffer from inferior translation quality due to removal of dependency on previous target tokens from input ...
作者: Xu, Manjie1; Jiang, Guangyuan2; Liang, Wei1, 3; Zhang, Chi4; Zhu, Yixin2 (1School of Computer Science & Technology, Beijing Institute of Technology, China;2Institute for AI, Peking University, China;3Yangtze Delta Region Academy, Beijing Institute of Technology, Jiaxing, China;4National Key Laboratory of General Artificial Intelligence, BIGAI, China)
出处: arXiv 2022
作者: 陶文彬1,2,3;钱育蓉1,2,3;张伊扬1,2,3;马恒志1,2,3;冷洪勇1,2,4;马梦楠1,2,3 (1新疆大学软件学院, 乌鲁木齐, 830046;2新疆大学软件学院重点实验室, 乌鲁木齐, 830046;3新疆维吾尔自治区信号检测与处理重点实验室, 新疆维吾尔自治区信号检测与处理重点实验室, 乌鲁木齐, 830046;4北京理工大学计算机学院, 北京, 100081)
出处: 计算机工程与应用 2022 第58卷 第18期 P16-25
关键词: 聚类算法;深度聚类;自编码器;特征提取
摘要: 聚类分析作为一种常见的分析方法,广泛应用于各种场景。随着机器学习技术的发展,深度聚类算法也成了当下研究的热点,基于自编码器的深度聚类算法是其中的代表算法。为了及时了解掌握基于自编码器的深度聚类算法的发展,介绍了四种自编码器的模型,对近些年代表性的算法依照自编码器的结构进行了分类。在MNIST、USP ...
作者: 张宇姣1;黄锐2;张福泉2;隋栋3;张虎4 (1太原师范学院教务处, 山西, 晋中, 030619;2北京理工大学计算机学院, 北京, 100081;3北京建筑大学电气与信息工程学院, 北京, 102406;4山西大学计算机与信息技术学院(大数据学院), 太原, 030006)
出处: 计算机科学 2022 第49卷 第5期 P165-169
关键词: 近邻传播;聚类;菌群优化;偏向参数
摘要: 为了提高近邻传播聚类算法的聚类性能,采用菌群算法进行近邻传播偏向参数优化求解。首先,根据待聚类样本建立相似矩阵,初始化偏向参数;然后采用菌群算法优化偏向参数,将偏向参数作为菌落进行训练,设置轮廓(Silhouette)指标值作为菌群算法的适应度函数;接着通过菌落位置更新优化后的偏向参数,进行近邻传播 ...
作者: Peng Cheng1;Zhang Chunxia1;Xue Xiaojun1;Gao Jiameng1;Liang Hongjian2;Niu Zhengdong1 (1School of Computer Science and Technology, Beijing Institute of Technology, Beijing, 100081;2School of Information, Production and Systems, Waseda University, Japan, Fukuoka, 808-0135)
出处: Tsinghua Science and Technology 2022 第27卷 第4期 P664-679
关键词: multimodal sentiment analysis;multimodal fusion;Cross-Modal Complementary Network (CMCN);hierarchical fusion;joint optimization
摘要: Multimodal Sentiment Classification (MSC) uses multimodal data, such as images and texts, to identify the users' sentiment polarities from the informa ...
作者: 魏恺轩;付莹 (北京理工大学计算机学院, 北京, 100081)
出处: 计算机科学 2022 第49卷 第8期 P120-126
关键词: 重参数化卷积单元;多尺度融合;空间通道并行注意力模块;极暗光图像降噪
摘要: 实用的暗光降噪增强解决方案往往需要具备计算速度快、内存效率高、能够实现视觉上高质量的降噪等优点。现有方法大多以提升降噪质量为目标,因此在速度和内存要求上有所折中,这在很大程度上限制了其实用性。文中提出了一种新的深度降噪网络---重参数化多尺度融合网络,用于极暗光单张原始图像降噪,在不损失降噪性能的同 ...
作者: Yang Jingsi;Huang Tianyu;Ding Gangyi;Li Lijie;Li Peng (School of Computer Science & Technology, Beijing Institute of Technology, Beijing, 100081)
出处: 系统仿真学报 2022 第34卷 第8期 P1750-1761
关键词: parallel simulation;modern live performance;multi-layer constraint hierarchy;whole process
摘要: A parallel simulation method is proposed for modern live performance. By decomposing the live performance process from the top down, this method assis ...