检索结果-内蒙古大学图书馆

38th Conference on Neural information processing Systems, NeurIPS 2024

作者： He, Xilin Hu, Jingyu Lin, Qinliang Luo, Cheng Xie, Weicheng Song, Siyang Khan, Muhammad Haris Shen, Linlin Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University China Shenzhen Institute of Artificial Intelligence and Robotics for Society China Guangdong Provincial Key Laboratory of Intelligent Information Processing China University of Exeter United Kingdom Mohamed bin Zayed University of Artificial Intelligence United Arab Emirates

Domain generalization methods aim to learn transferable knowledge from source domains that can generalize well to unseen target domains. Recent studies show that neural networks frequently suffer from a simplicity-biased learning behavior which leads to over-reliance on specific frequency sets, namely as frequency shortcuts, instead of semantic information, resulting in poor generalization performance. Despite previous data augmentation techniques successfully enhancing generalization performances, they intend to apply more frequency shortcuts, thereby causing hallucinations of generalization improvement. In this paper, we aim to prevent such learning behavior of applying frequency shortcuts from a data-driven perspective. Given the theoretical justification of models' biased learning behavior on different spatial frequency components, which is based on the dataset frequency properties, we argue that the learning behavior on various frequency components could be manipulated by changing the dataset statistical structure in the Fourier domain. Intuitively, as frequency shortcuts are hidden in the dominant and highly dependent frequencies of dataset structure, dynamically perturbating the over-reliance frequency components could prevent the application of frequency shortcuts. To this end, we propose two effective data augmentation modules designed to collaboratively and adaptively adjust the frequency characteristic of the dataset, aiming to dynamically influence the learning behavior of the model and ultimately serving as a strategy to mitigate shortcut learning. Code is available at AdvFrequency. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Clustering Algorithm with Learnable Distance for Categorical Data with Nominal and Ordinal Attributes 5

Clustering Algorithm with Learnable Distance for Categorical...

引用

5th International Conference on Data Science and information Technology, DSIT 2022

作者： Jia, Hong Zhong, Weiwei College of Electronics and Information Engineering Guangdong Key Laboratory of Intelligent Information Processing Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen Key Laboratory of Media Security Shenzhen University Shenzhen China

ISBN: (数字)9781665498685

ISBN: (纸本)9781665498685

For clustering analysis on categorical data, the distance measurement between two objects often plays a very important role. However, most of the existing categorical distance metrics do not distinguish between nominal attributes and ordinal attributes. That is, these metrics do not explore the information contained in ordinal values, and ignore the order relationship among them. Therefore, this paper proposes a novel clustering algorithm, which uses a united framework to measure the distance between nominal attributes and ordinal attributes while distinguishing the different characteristics between them. The basic idea of the proposed method is that, the attribute value pairs with larger co-occurrence probability in the same cluster may have smaller distances. Therefore, the distances between different categories are dynamically evaluated based on the current cluster structure of the data samples. Subsequently, the distances and cluster relationship are alternately learned until convergence. Experimental results show that the proposed algorithm has better robustness and performance than the existing counterparts on different kinds of categorical data sets. © 2022 IEEE.

关键词： Clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

Super-resolution reconstruction algorithm for terahertz imaging below diffraction limit

引用

Chinese Physics B 2023年第3期32卷 608-612页

作者：王莹祁峰张子旭汪晋宽 School of Communication Science and Engineering Northeastern UniversityShenyang 110819China Key Laboratory of Opto-Electronic Information Processing Chinese Academy of SciencesShenyang 110016China Shenyang Institute of Automation Chinese Academy of SciencesShenyang 110016China Institutes for Robotics and Intelligent Manufacturing Chinese Academy of SciencesShenyang 110169China University of Chinese Academy of Sciences Beijing 100049China University of Technology Sydney NSW 2007Australia

Terahertz(THz)imaging has drawn significant attention because THz wave has a unique capability to transient,ultrawide spectrum and low photon ***,the low resolution has always been a problem due to its long wavelength,limiting their application of fields practical *** this paper,we proposed a complex one-shot super-resolution(COSSR)framework based on a complex convolution neural network to restore superior THz images at 0.35 times wavelength by extracting features directly from a reference measured sample and groundtruth without the measured *** with real convolution neural network-based approaches and complex zero-shot super-resolution(CZSSR),COSSR delivers at least 6.67,0.003,and 6.96%superior higher imaging efficacy in terms of peak signal to noise ratio(PSNR),mean square error(MSE),and structural similarity index measure(SSIM),respectively,for the analyzed ***,the proposed method is experimentally demonstrated to have a good generalization and to perform well on measured *** COSSR provides a new pathway for THz imaging super-resolution(SR)reconstruction below the diffraction limit.

关键词： terahertz image processing complex convolution neural network

来源：评论

学校读者我要写书评

暂无评论

Learning to Localize Cross-Anatomy Landmarks in X-Ray Images with a Universal Model

引用

Biomedical Engineering Frontiers 2022年第1期3卷 298-308页

作者： Heqin Zhu Qingsong Yao Li Xiao S.Kevin Zhou Key Lab of Intelligent Information Processing of Chinese Academy of Sciences(CAS) Institute of Computing TechnologyCASBeijing 100190China Center for Medical Imaging RoboticsAnalytic Computing&Learning(MIRACLE)School of Biomedical Engineering&Suzhou Institute for Advanced ResearchUniversity of Science and Technology of ChinaSuzhou 215123China

Objective and Impact *** this work,we develop a universal anatomical landmark detection model which learns once from multiple datasets corresponding to different anatomical *** with the conventional model trained on a single dataset,this universal model not only is more light weighted and easier to train but also improves the accuracy of the anatomical landmark *** accurate and automatic localization of anatomical landmarks plays an essential role in medical image ***,recent deep learning-based methods only utilize limited data from a single *** is promising and desirable to build a model learned from different regions which harnesses the power of big *** model consists of a local network and a global network,which capture local features and global features,*** local network is a fully convolutional network built up with depth-wise separable convolutions,and the global network uses dilated convolution to enlarge the receptive field to model global *** evaluate our model on four 2D X-ray image datasets totaling 1710 images and 72 landmarks in four anatomical *** experimental results show that our model improves the detection accuracy compared to the state-of-the-art *** model makes the first attempt to train a single network on multiple datasets for landmark *** results qualitatively and quantitatively show that our proposed model performs better than other models trained on multiple datasets and even better than models trained on a single dataset separately.

关键词： convolution utilize separable

来源：评论

学校读者我要写书评

暂无评论

A single-stream adaptive scene layout modeling method for scene recognition

引用

Neural Computing and Applications 2024年第22期36卷 13703-13714页

作者： Wang, Qun Zhu, Feng Lin, Zhiyuan Wang, Jianyu Li, Xiang Zhao, Pengfei Key Laboratory of Opto-Electronic Information Processing Chinese Academy of Sciences Shenyang110016 China Shenyang Institute of Automation Chinese Academy of Sciences Shenyang110016 China Institutes for Robotics and Intelligent Manufacturing Chinese Academy of Sciences Shenyang110016 China University of Chinese Academy of Sciences Beijing100049 China Faculty of Robot Science and Engineering Northeastern University Shenyang110016 China Faculty of Computer Science and Engineering Northeastern University Shenyang110016 China

Scene recognition has been the foundation of research in computer vision fields. Because scene images typically are composed of specific regions distributed in some layout, so modeling layouts of various scenes is a key clue for scene recognition. Existing methods usually require an additional stream to detect regions for subsequent modeling, which accumulate errors and may miss important information. Meanwhile, they use manual features to model relations between regions, which weakens the representation ability of layouts. In this paper, we propose a single-stream adaptive scene layout modeling approach based on a layout modeling module (LMM), which constructs layouts without additional detection streams and adaptively captures the relations to take advantage of graph attention network. LMM is directly concatenated to a convolutional neural network, where each pixel of the activation maps of the last convolutional layer is defined as a region that is the initial input node of the LMM. LMM first models the layout of each region, and then uses all regions with layout information to model the entire scene. Layout relations are encoded as edges, which are automatically analyzed according to region co-occurrence and relative position. Our work can be understood as optimizing features of the activation maps from a scene layout modeling perspective for scene recognition. Experimental results on MIT67, SUN397, and Places365 show that our single-stream model achieves competitive performance. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

CB-YOLOv5 Algorithm for Small Target Detection in Aerial Images 4

CB-YOLOv5 Algorithm for Small Target Detection in Aerial Ima...

引用

4th International Conference for Emerging Technology, INCET 2023

作者： Li, Yingjie Wang, Yitian Zhao, Huaici Shenyang University of Chemical Technology College of Information Engineering Shenyang110142 China Chinese Academy of Sciences Key Laboratory of Opto-Electronic Information Processing Shenyang110016 China Chinese Academy of Sciences Shenyang Institute of Automation Shenyang110016 China Chinese Academy of Sciences Institutes for Robotics and Intelligent Manufacturing Shenyang110169 China

ISBN: (纸本)9798350335750

Aerial images are often plagued by background interference, and small targets with indistinct features, leading to low accuracy, high false detection rates, and high miss detection rates. To address these challenges, a small target detection algorithm based on YOLOv5, Coordinate-attention and Bidirectional-feature-pyramid-network YOLOv5 (CB-YOLOv5), is proposed in this paper. Considering the small number of pixels occupied by small targets and their indistinct features, a fourth target detection layer is added by concatenating the feature map from quadruple down-sampling during feature extraction with the feature map output from 8-fold up-sampling during feature fusion. Additionally, a coordinate attention mechanism is introduced during the feature extraction stage to improve small target localization and enhance detection accuracy. Finally, the original Path Aggregation Networks (PANet) structure is replaced with a weighted Bidirectional Feature Pyramid Network (BiFPN) structure during the feature fusion stage to improve the network's ability to fuse feature maps of different scales. The simulation results demonstrate that the CB-YOLOv5 improves mAP50 by 9.4%, mAP75 by 9.7%, and mAP50:95 by 7.8% compared to the original YOLOv5s model. Thus, the effectiveness of the CB-YOLOv5 algorithm for detecting small targets in aerial images is validated. © 2023 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Augmented Feature Representation with Parallel Convolution for Cross-domain Facial Expression Recognition 16th

Augmented Feature Representation with Parallel Convolution ...

引用

16th Chinese Conference on Biometric Recognition, CCBR 2022

作者： Yang, Fan Xie, Weicheng Zhong, Tao Hu, Jingyu Shen, Linlin Computer Vision Institute School of Computer Science and Software Engineering Shenzhen University Shenzhen China Shenzhen Institute of Artificial Intelligence and Robotics for Society Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen China

ISBN: (纸本)9783031202322

Facial expression recognition (FER) has made significant progress in the past decade, but the inconsistency of distribution between different datasets greatly limits the generalization performance of a learned model on unseen datasets. Recent works resort to aligning feature distributions between domains to improve the cross-domain recognition performance. However, current algorithms use one output each layer for the feature representation, which can not well represent the complex correlation among multi-scale features. To this end, this work proposes a parallel convolution to augment the representation ability of each layer, and introduces an orthogonal regularization to make each convolution represent independent semantic. With the assistance of a self-attention mechanism, the proposed algorithm can generate multiple combinations of multi-scale features to allow the network to better capture the correlation among the outputs of different layers. The proposed algorithm achieves state-of-the-art (SOTA) performances in terms of the average generalization performance on the task of cross-database (CD)-FER. Meanwhile, when AFED or RAF-DB is used for the training, and other four databases, i.e. JAFFE, SFEW, FER2013 and EXPW are used for testing, the proposed algorithm outperforms the baselines by the margins of 5.93% and 2.24% in terms of the average accuracy. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

ExpTopo: Explicit Topological Modeling for Bronchus Segmentation 22

ExpTopo: Explicit Topological Modeling for Bronchus Segmenta...

引用

22nd IEEE International Symposium on Biomedical Imaging, ISBI 2025

作者： Zhao, Mingyue Qiu, Xiaolan Zhou, S.Kevin School of Biomedical Engineering Division of Life Sciences and Medicine University of Science and Technology of China (USTC) China Center for Medical Imaging Robotics Analytic Computing & Learning (MIRACLE) Suzhou Institute for Advance Research Ustc China Suzhou Key Laboratory of Microwave Imaging Processing and Application Technology Suzhou Aerospace Information Research Institute China Key Laboratory of Precision and Intelligent Chemistry University of Science and Technology of China (USTC) China Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS) Institute of Computing Technology China

ISBN: (纸本)9798331520526

Airway extraction is paramount in the early diagnosis and treatment of respiratory diseases. As a tree-like structure, both topological-aware learning and voxel-wise classification are equally crucial for the airway. However, existing methods demonstrate insufficient topological learning, emphasizing only the supervision of individual key topological points. Consequently, this paper proposes a Explicit Topological Modeling (ExpTopo) approach to aid airway segmentation. It explicitly introduces topological metric space learning based on semantic segmentation, enhancing the model's structural perception by implementing global skeleton-level sparse topological learning (STL) and local voxel-level dense topological perception (DTP). Extensive experimental results demonstrate that the algorithm achieves competitive performance at both the topological and voxel levels. Code will be available in https://***/MorineZ/ExpTopo. © 2025 IEEE.

关键词： Airway extraction dense topological perception explicit topological modeling sparse topological learning

来源：评论

学校读者我要写书评

暂无评论

FLIP-80M: 80 Million Visual-Linguistic Pairs for Facial Language-Image Pre-Training 24

FLIP-80M: 80 Million Visual-Linguistic Pairs for Facial Lang...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Li, Yudong Hou, Xianxu Dezhi, Zheng Shen, Linlin Zhao, Zhe School of Computer Science and Software Engineering Shenzhen University Shenzhen China Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen China School of AI and Advanced Computing Xi'an Jiaotong-Liverpool University Shenzhen China Guangdong Provincial Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen China Tencent AI Lab Beijing China

ISBN: (纸本)9798400706868

While significant progress has been made in multi-modal learning driven by large-scale image-text datasets, there is still a noticeable gap in the availability of such datasets within the facial domain. To facilitate and advance the field of facial representation learning, we present FLIP-80M, a large-scale visual-linguistic dataset comprising over 80 million face images paired with text descriptions. FLIP-80M is constructed by leveraging the large openly available image-text-pair dataset LAION-5B and a mixed-method approach to filter face-related pairs from both visual and linguistic perspectives. Our curation process involves face detection, face caption classification, text de-noising, and synthesis-based image augmentation. As a result, FLIP-80M stands as the largest face-text dataset to date. To evaluate the potential of our dataset, we fine-tune the CLIP model using the proposed FLIP-80M, to create FLIP (Facial Language-Image Pretraining) and assess its representation capabilities across various downstream tasks. Our experiments demonstrate that our FLIP model achieves state-of-the-art results in a range of face analysis tasks, including face parsing, face alignment, and face attribute classification. The dataset and models are available at https://***/ydli-ai/FLIP. © 2024 ACM.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Spectral Ensemble Clustering with LDA-based Co-training for Multi-view Data Analysis 17

Spectral Ensemble Clustering with LDA-based Co-training for ...

引用

17th International Conference on Computational Intelligence and Security, CIS 2021

作者： Jia, Hong Li, Zhimin Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen University Guangdong Key Laboratory of Intelligent Information Processing Shenzhen China

ISBN: (纸本)9781665494892

Multi-view data has attracted more and more attention in recent years due to its practical significance. Generally, ensemble clustering and co-training are effective ways to analyze this kind of data. Among ensemble clustering algorithms, Spectral Ensemble Clustering (SEC) shows good performance by leveraging co-association matrix, which is constructed with base clustering result of single-view space. However, SEC is also limited by base clustering results. When the base clustering results of some views deviate greatly from the actual data distribution, it will affect the accuracy of co-association matrix, making it difficult for SEC to learn good cluster structure. Therefore, this paper proposes a novel algorithm for multi-view data clustering, namely Spectral Ensemble Clustering with LDA-based Co-training (LSEC). This method introduces a novel co-training strategy into ensemble clustering analysis. Specifically, based on the assumption that the clustering result of SEC has global cluster information, we use the class label obtained by SEC as reference label to perform Linear Discriminant Analysis (LDA) on original data to get refined data. Subsequently, K-means algorithm is performed on the refined data to get new base clustering results. The basic idea of this strategy is to use global cluster information to co-train local feature space. Experimental results on seven real world multi-view data sets show the effectiveness of proposed method. © 2021 IEEE.

关键词： K-means clustering

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：