检索结果-内蒙古大学图书馆

2011 International Conference on Automation and Robotics, ICAR 2011

作者： Gao, Chaobang Zhou, Jiliu Lang, Fangnian Pu, Qiang Liu, Chang College of Computer Science and Technology Chengdu University Chengdu 610106 China School of Computer Science Sichuan University Chengdu 610064 China Key Laboratory of Pattern Recognition and Intelligent Information Processing Sichuan Province Chengdu 610106 China

ISBN: (纸本)9783642255526

In this paper, we denote a color image by a quaternion function, then find edge points by solving the maximum of quaternion fractional directional differentiation(QFDD)'s norm. This method is called edge detection based on QFDD. Experiments indicate that the method has special advantages. Comparing with Canny, LOG, Sobel, and general fractional differentiation, we discover that QFDD has fewer false negatives in the textured regions and is also better at detecting edges which are partially defined by texture, which means we will obtain better results in the interesting regions by QFDD and these results are more consistent with the characteristics of human visual system. © 2011 Springer-Verlag.

关键词： Edge detection

来源：评论

学校读者我要写书评

暂无评论

Dynamic Pareto Optimal Matching

Dynamic Pareto Optimal Matching

引用

International Symposium on information Science and Engineering, ISISE

作者： Rudolf Fleischer Yihui Wang Department of Computer Science Shanghai Key Laboratory of Intelligent Information Processing Fudan University Shanghai China

We consider the problem of assigning houses to agents, where each agent has his own partial ranking of the houses (i.e., a preference list of a subset of the houses). A matching M is Pareto optimal if there exists no other matching M' such that all agents have a house in M' not worse than in M and at least one agent has a better house in M'. In a dynamic market where agents and houses can be added or removed at any time, a Pareto optimal matching may lose its optimality. In this paper, we give an O(m) time algorithm to maintain a maximum cardinality and Pareto optimal matching, where m is the total size of all preference lists. Furthermore, we show that a Pareto optimal matching can be transformed to any other Pareto optimal matching through a certain constructed graph in linear time.

关键词： Optimal matching Stability Switches information science Computer science Laboratories intelligent agent information processing Educational institutions Voting

来源：评论

学校读者我要写书评

暂无评论

A Generative Framework for Self-Supervised Facial Representation Learning

arXiv

引用

arXiv 2023年

作者： He, Ruian Xing, Zhen Tan, Weimin Yan, Bo School of Computer Science Shanghai Key Laboratory of Intelligent Information Processing Fudan University Shanghai China

Self-supervised representation learning has gained increasing attention for strong generalization ability without relying on paired datasets. However, it has not been explored sufficiently for facial representation. Self-supervised facial representation learning remains unsolved due to the coupling of facial identities, expressions, and external factors like pose and light. Prior methods primarily focus on contrastive learning and pixel-level consistency, leading to limited interpretability and suboptimal performance. In this paper, we propose LatentFace, a novel generative framework for self-supervised facial representations. We suggest that the disentangling problem can be also formulated as generative objectives in space and time, and propose the solution using a 3D-aware latent diffusion model. First, we introduce a 3D-aware autoencoder to encode face images into 3D latent embeddings. Second, we propose a novel representation diffusion model to disentangle 3D latent into facial identity and expression. Consequently, our method achieves state-of-the-art performance in facial expression recognition (FER) and face verification among self-supervised facial representation learning models. Our model achieves a 3.75% advantage in FER accuracy on RAF-DB and 3.35% on AffectNet compared to SOTA methods. Copyright © 2023, The Authors. All rights reserved.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

CT2C-QA: Multimodal Question Answering over Chinese Text, Table and Chart 24

CT2C-QA: Multimodal Question Answering over Chinese Text, Ta...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Zhao, Bowen Cheng, Tianhao Zhang, Yuejie Cheng, Ying Feng, Rui Zhang, Xiaobo Fudan University Shanghai China Children's Hospital of Fudan University Shanghai China School of Computer Science Shanghai Key Laboratory of Intelligent Information Processing Fudan University China Shanghai Collaborative Innovation Center of Intelligent Visual Computing. China Fudan Zhangjiang Institute Shanghai China Children's Hospital of Fudan University National Children's Medical Center Shanghai China

ISBN: (纸本)9798400706868

Multimodal Question Answering (MMQA) is crucial as it enables comprehensive understanding and accurate responses by integrating insights from diverse data representations such as tables, charts, and text. Most existing researches in MMQA only focus on two modalities such as image-text QA, table-text QA and chart-text QA, and there remains a notable scarcity in studies that investigate the joint analysis of text, tables, and charts. In this paper, we present CT2C-QA, a pioneering Chinese reasoning-based QA dataset that includes an extensive collection of text, tables, and charts, meticulously compiled from 200 selectively sourced webpages. Our dataset simulates real webpages and serves as a great test for the capability of the model to analyze and reason with multimodal data, because the answer to a question could appear in various modalities, or even potentially not exist at all. Additionally, we present AED (Allocating, Expert and Decision), a multi-agent system implemented through collaborative deployment, information interaction, and collective decision-making among different agents. Specifically, the Assignment Agent is in charge of selecting and activating expert agents, including those proficient in text, tables, and charts. The Decision Agent bears the responsibility of delivering the final verdict, drawing upon the analytical insights provided by these expert agents. We execute a comprehensive analysis, comparing AED with various state-of-the-art models in MMQA, including GPT-4. The experimental outcomes demonstrate that current methodologies, including GPT-4, are yet to meet the benchmarks set by our dataset. © 2024 ACM.

关键词： Question answering

来源：评论

学校读者我要写书评

暂无评论

Pose Optimization of a Single Ball Self-balancing Mobile Robot Based on IMU and Visual Fusion

Pose Optimization of a Single Ball Self-balancing Mobile Rob...

引用

第37届中国控制会议

作者： She Jing Zhongli Ma Kang Yi Huixin Li Zuoyong Li Harbin Engineering University Fujian Provincial Key Laboratory of Information Processing and Intelligent Control Minjiang University

The single ball self-balancing mobile robot is a typical multivariable, high-order, nonlinear and static instability system, IMU inertial devices are generally used to obtain attitude parameters of the system. The data errors obtained by the inertial devices will accumulate over time, and the long-term pose estimation is rather unreliable. But for some quick motion over a short period of time, inertial sensors can get very accurate estimates. Monocular cameras are often used as image real-time monitoring sensors by single ball self-balancing mobile robots. The camera’s data does not accumulate errors over time, so there is essentially no drift. Therefore, camera pose estimation can effectively correct the inertial sensor error. In a fast-moving environment, the shutter-roller camera cannot capture the image clearly and there may be ghosting or overlapping of two areas too little, resulting in mismatched or unmatched conditions. So that the robot makes a wrong estimate. In this paper, the non-linear optimization method is used to fuse the IMU and monocular attitude data to obtain the optimal pose estimation of the robot. Based on the modeling of robot using Lagrange’s equation method, the control of single ball robot is realized by using fuzzy PID algorithm. Experiments show that this scheme can respond quickly to the change of the attitude of the ball self-balancing robot and keep the attitude of the robot relatively stationary, so that the robot can move steadily.

关键词： Ball self-balancing robot Attitude detection Nonlinear optimization Visual inertial data fusion

来源：评论

学校读者我要写书评

暂无评论

From Quick-draw To Story: A Story Generation System for Kids’ Robot

From Quick-draw To Story: A Story Generation System for Kids...

引用

IEEE International Conference on Robotics and Biomimetics

作者： Lecheng Wang Shizheng Qin Menglong Xu Rui Zhang Lizhe Qi Wenqiang Zhang Shanghai Key Laboratory of Intelligent Information Processing Fudan University Shanghai P.R. China

From quick-draw to story aims to draw a narrative story based on simple lines children draw. On account of the simple knowledge contained in quick-draw and lack of labelled data, this task is assuredly challenging. In this paper, it is divided into two subtasks. Firstly, a Multitask Transformer Network is proposed to generate the sentence on the grounds of the information of quick-draw. Secondly, we finetune the OpenAI Generative Pre-training Model (OpenAI GPT) to generate a story on the basis of the sentence. However, evaluating the text generation model is difficult. In order to solve with this, a new evaluation metric based on Story Cloze Test is proposed. Our evaluations indicate the prominent results our method has achieved in this task.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prediction of Drug-Target Interactions with CNNs and Random Forest 16th

Prediction of Drug-Target Interactions with CNNs and Random ...

引用

16th International Conference on intelligent computing, ICIC 2020

作者： Lin, Xiaoli Xu, Minqi Yu, Haiping Hubei Key Laboratory of Intelligent Information Processing and Real-Time Industrial System School of Computer Science and Technology Wuhan University of Science and Technology Wuhan430065 China School of Computer Huanggang Normal University Huanggang438000 China

ISBN: (纸本)9783030608019

Drug molecules interact with target proteins to influence the pharmacological action of the target to achieve the phenotypic effect, which can facilitate the identification of novel targets for current drug. Traditional biological experiments for discovering new drug-target interactions are expensive and time-consuming. Therefore, it is crucial to develop new prediction methods for identifying potential drug-target interactions. computing methods have been increasing developed which can quickly and effectively predict drug-target interactions. In particular, machine learning methods have been widely used due to high predictive performance and computational efficiency. This paper first uses MACCS substructure fingerings to encode the drug molecules, then uses CNNs to extract the biological evolutionary information of target protein sequences, and finally uses random forest algorithm to predict drug-target interactions. Four datasets of drug-target interactions including Enzymes, Ion Channels, GPCRs and Nuclear Receptors, are independently used for building models with random forest. The results demonstrate our proposed method has a general compatibility, which is effective and feasible to predict drug-target interactions. © 2020, Springer Nature Switzerland AG.

关键词： Decision trees

来源：评论

学校读者我要写书评

暂无评论

A novel neural network classification model based on covering and Affinity Propagation clustering algorithm

引用

Journal of Computational information Systems 2013年第7期9卷 2565-2573页

作者： Li, Hui Ding, Shifei School of Computer Science and Technology China University of Mining and Technology Xuzhou 221116 China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing 100190 China Jiangsu Normal University Xuzhou 221116 China

Affinity Propagation (AP) algorithm can automatically determine the cluster center and does not need a pre-determined number of clustering. This paper presents a novel neural network classification model, using AP clustering method as a pre-processor to cluster high dimensional massive data samples, and combining covering algorithm to automatically determine the center of the covering, and calculating its radius, and then converting these parameters into the weights and thresholds of the hidden layer of classification neural network. The number of hidden layer neurons is the number of covering. The output layer is the M-P neuron layer. By evaluating the input feature of hidden layer to complete the mapping from covering to the output, and gain the corresponding output category of the input feature. Simulation results show that our proposed neural network classification model can get better performance. Copyright © 2013 Binary information Press.

关键词： Clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

Solving Constrained Multi-objective Optimization Problems with Passive Archiving Mechanism 1

引用

17th International Conference on Bio-Inspired computing: Theories and Applications, BIC-TA 2022

作者： Jia, Huijuan Zhang, Kai Shen, Chaonan School of Computer Science and Technology Wuhan University of Science and Technology Wuhan430065 China Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System Wuhan University of Science and Technology Wuhan430065 China

ISBN: (数字)9789819915491

ISBN: (纸本)9789819915484

During these years, many advanced constrained multi-objective evolutionary algorithms (CMOEAs) have been developed to solve constrained multi-objective optimization problems (CMOPs). However, some existing constrained multi-objective algorithms do not use an archiving mechanism, which leads to easy loss of the searched feasible solutions during the search process. In addition, even if some algorithms use an archiving mechanism, an unsuitable archiving update strategy can cause difficulty in balancing the diversity and convergence of the archive set. To address these challenges, a two-stage constrained multi-objective evolutionary algorithm based on a passive archiving mechanism (PA-CMOEA) is proposed in this paper. In the first stage, the constrained multi-objective problem is modeled as an unconstrained multi-objective problem, which allows the population to fully explore the decision space. The archive set is passively updated in a constraint-domination principle (CDP) criterion using population information. The goal of this stage is for the archive set to converge to the constrained Pareto front region of the problem with the help of population information. In the second stage, a local search is performed on the archive set. Specifically, the convergence and diversity of the archived solution set is improved while ensuring that the feasibility of the archived set is not worse. Finally, the algorithm outputs the archive set. To estimate the performance of the proposed algorithm, experiments are conducted on the LIRCMOP and CF benchmarks in comparison with some of the most popular algorithms. The experimental results demonstrate that the proposed algorithm has a competitive advantage compared with the comparative algorithms. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Adaptive nonlinear auto-associative modeling through manifold learning

Adaptive nonlinear auto-associative modeling through manifol...

引用

9th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2005

作者： Zhang, Junping Li, Stan Z. Intelligent Information Processing Laboratory Department of Computer Science and Engineering Fudan University Shanghai 200433 China Key Laboratory of Complex Systems and Intelligence Science Chinese Academy of Sciences China National Laboratory of Pattern Recognition Center for Biometrics and Security Research Institute of Automation Beijing 100080 China

ISBN: (纸本)3540260765

We propose adaptive nonlinear auto-associative modeling (ANAM) based on Locally Linear Embedding algorithm (LLE) for learning intrinsic principal features of each concept separately and recognition thereby. Unlike traditional supervised manifold learning algorithm, the proposed ANAM algorithm has several advantages: 1) it implicitly embodies discriminant information because the suboptimal parameters of ANAM are determined based on error rate of the validation set. 2) it avoids the curse of dimensionality without loss accuracy because recognition is completed in the original space. Experiments on character and digit databases show that the advantages of the proposed ANAM algorithm. © Springer-Verlag Berlin Heidelberg 2005.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：