检索结果-内蒙古大学图书馆

Rethinking the Sample Relations for Few-Shot Classification

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Yin, Guowei Huang, Sheng Huangfu, Luwen Zhang, Yi Zhang, Xiaohong School of Big Data and Software Engineering Chongqing University No.55 Daxuecheng South Rd. Shapingba Chongqing401331 China Ministry of Education Key Laboratory of Dependable Service Computing in Cyber Physical Society Chongqing University No.174 Shazhengjie Shapingba Chongqing400044 China Fowler College of Business San Diego State University San DiegoCA92182 United States San Diego State University San DiegoCA92182 United States AI4Business Lab San Diego State University San DiegoCA92182 United States

Feature quality is paramount for classification performance, particularly in few-shot scenarios. Contrastive learning, a widely adopted technique for enhancing feature quality, leverages sample relations to extract intrinsic features that capture semantic information and has achieved remarkable success in Few-Shot Learning (FSL). Nevertheless, current few-shot contrastive learning approaches often overlook the semantic similarity discrepancies at different granularities when employing the same modeling approach for different sample relations, which limits the potential of few-shot contrastive learning. In this paper, we introduce a straightforward yet effective contrastive learning approach, Multi-Grained Relation Contrastive Learning (MGRCL), as a pretraining feature learning model to boost few-shot learning by meticulously modeling sample relations at different granularities. MGRCL categorizes sample relations into three types: intra-sample relation of the same sample under different transformations, intra-class relation of homogenous samples, and inter-class relation of inhomogeneous samples. In MGRCL, we design Transformation Consistency Learning (TCL) to ensure the rigorous semantic consistency of a sample under different transformations by aligning predictions of input pairs. Furthermore, to preserve discriminative information, we employ Class Contrastive Learning (CCL) to ensure that a sample is always closer to its homogenous samples than its inhomogeneous ones, as homogenous samples share similar semantic content while inhomogeneous samples have different semantic content. Our method is assessed across four popular FSL benchmarks, showing that such a simple pre-training feature learning method surpasses a majority of leading FSL methods. Moreover, our method can be incorporated into other FSL methods as the pre-trained model and help them obtain significant performance gains. © 2025, CC BY-NC-SA.

关键词： Contrastive Learning

Abnormality Detection in Chest X-Ray Images Using Uncertainty Prediction Autoencoders 23rd

学校读者我要写书评

暂无评论

Abnormality Detection in Chest X-Ray Images Using Uncertaint...

23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020

作者： Mao, Yifan Xue, Fei-Fei Wang, Ruixuan Zhang, Jianguo Zheng, Wei-Shi Liu, Hongmei School of Data and Computer Science Sun Yat-sen University Guangzhou China Key Laboratory of Machine Intelligence and Advanced Computing MOE Guangzhou China Department of Computer Science and Engineering Southern University of Science and Technology Shenzhen China Pazhou Lab Guangzhou China Guangdong Province Key Laboratory of Information Security Technology Guangzhou China

ISBN: (纸本)9783030597245

Chest radiography is widely used in annual medical screening to check whether lungs are healthy or not. Therefore it would be desirable to develop an intelligent system to help clinicians automatically detect potential abnormalities in chest X-ray images. Here with only healthy X-ray images, we propose a new abnormality detection approach based on an autoencoder which outputs not only the reconstructed normal version of the input image but also a pixel-wise uncertainty prediction. Higher uncertainty often appears at normal region boundaries with relatively larger reconstruction errors, but not at potential abnormal regions in the lung area. Therefore the normalized reconstruction error by the uncertainty provides a natural measurement for abnormality detection in images. Experiments on two chest X-ray datasets show the state-of-the-art performance by the proposed approach. © 2020, Springer Nature Switzerland AG.

关键词： Forecasting

Multi-Scale Convolution-Transformer Fusion Network for Endoscopic Image Segmentation

学校读者我要写书评

暂无评论

Multi-Scale Convolution-Transformer Fusion Network for Endos...

IEEE International Symposium on Biomedical Imaging

作者： Baosheng Zou Zongguang Zhou Ying Han Kang Li Guotai Wang West China Biomedical Big Data Center Sichuan University Chengdu China School of Mechanical and Electrical Engineering University of Electronic Science and Technology of China Chengdu China Department of General Surgery Division of Gastrointestinal Surgery West China Hospital of Sichuan University Chengdu China Institute of Digestive Surgery & State Key Laboratory of Biotherapy Sichuan University Chengdu China West China Medical Simulation Center Sichuan University West China Hospital Chengdu China Shanghai AI Lab Shanghai China

Automatic semantic segmentation of endoscopic images is an essential part of computer-assisted intervention surgery. Recently, Convolutional Neural Networks (CNNs) have been widely applied to endoscopic image segmentation, but their performance is still limited due to the weak ability to capture global long-range dependencies. This paper proposes a model that combines CNN and Transformer to deal with this problem, and it is named as Multi-scale Convolution-Transformer Fusion Network (MCTFNet) and consists of three components: 1) Multiple-parallel Multi-scale Transformer Convolution (MMTC) modules in parallel branches to extract Multi-scale information, 2) Multi-scale Information Fusion (MIF) module that fuses parallel branch information to allow interaction between different resolutions and 3) High-resolution Information Processing (HIP) module to keep high-resolution features in the image and avoid loss of details. We verified our method on HeiSurF dataset, and the results show that our method achieved an average Dice of 80.07%, which outperformed state-of-the-art CNNs including HRNet (79.93%) and Deeplabv3 (78.34%). It also outperformed several networks designed for medical image segmentation.

关键词：

Research on DNS Resolution Technology based on Blockchain

学校读者我要写书评

暂无评论

Research on DNS Resolution Technology based on Blockchain

Chinese Automation Congress (CAC)

作者： Wang Hu-qing Zhang Ying-jiao Zhang Xu Sun Zhi-xin Post Industry Technology Research and Development Center of the State Posts Bureau(Intemet of Things Technology) Post Big Data Technology and Application Engineering Research Center of Jiangsu Province Nanjing University of Posts and Telecommunications Key Lab of Broadband Wireless Communication and Sensor Network Technology Ministry of Education Nanjing University of Posts and Telecommunications Nanjing China School of Modern Posts Nanjing University of Posts and Telecommunications Nanjing China Aahui Yougu Express Intelligent Technology Co. LTD Nanjing China

DNS completes the resolution of domain names to IP addresses, which is an effective means to realize various In-ternet applications, and is also an indispensable part of Internet infrastructure, playing an important role. However, the resolution efficiency and security of DNS have always been a hot spot in the industry. This study deeply discusses the principle of Blockchain-based DNS resolution and conducts extensive re-search on Blockchain-based DNS resolution schemes. Based on compatibility with traditional DNS, these schemes are divided into two categories: compatible and incompatible. For different scenarios, we conduct a detailed analysis and comparison, focusing on the technical characteristics and performance of each so-lution. Finally, the future research direction is looked forward to, and the full text is summarized.

关键词：

Dual Autoencoder-Based Joint Learning: Enhancing Depth Classification of Hard Inclusions in Soft Tissue for Robotic Palpation

学校读者我要写书评

暂无评论

IEEE/ASME Transactions on Mechatronics 2025年

作者： Wang, Jingnan Zhou, Zhenning Cui, Yunduan Cui, Jian Qin, Longhui Xu, Tiantian Yi, Zhengkun Wu, Xinyu Chinese Academy of Sciences Guangdong Provincial Key Laboratory of Robotics and Intelligent System Shenzhen Institutes of Advanced Technology Shenzhen 518055 China Shenzhen University of Advanced Technology Center for Neurocognition and Social Behavior Institute of Artificial Intelligence Shenzhen 518107 China Research Center for Augmented Intelligence Research Institute of Artificial Intelligence Zhejiang Lab Zhejiang 310000 China Southeast University School of Mechanical Engineering Nanjing 211189 China City University of Macau Faculty of Data Science 999078 Macao Key Laboratory of Biomedical Imaging Science and System Chinese Academy of Sciences Beijing 100864 China Shenzhen University of Advanced Technology Shenzhen 518107 China

In robot-assisted minimally invasive surgery (RMIS), robotic palpation is vital for enhancing tissue assessment accuracy, especially for tumor depth detection, which is crucial for precise resections and improved treatment outcomes. By complementing and leveraging the strengths of data from multiple tactile sensors, the tumor detection task can realize significant improvements in accuracy and reliability, thereby enhancing overall performance in robotic palpation in RMIS. However, challenges arise due to differences in sensor modalities and the lack of unified data representation. To address this, we fabricate silicone-based phantom tissue to simulate soft tissue and embedded simulated tumors with hard inclusions at depths ranging from 0–11 mm, treating it as a 12-class classification problem. We conduct two robotic palpation experiments: one with the BarrettHand capacitive sensor and the other with the Digit sensor, collecting two tactile datasets. To explore the joint learnability of datasets collected from different tactile sensors, we design a dual autoencoder-based joint learning framework that integrates two recurrent autoencoders to process the two different tactile datasets. By applying a joint loss mechanism to connect their latent spaces, the autoencoders are trained jointly together, with the latent representation used for supervised classification. Extensive experiments show that joint learning enables sharing of features learned from different tactile datasets, thereby enhancing learning efficiency and classification accuracy for tactile datasets with different modalities and improving the performance of both autoencoders compared to independent training. © 1996-2012 IEEE.

关键词： Classification task dual autoencoder joint learning robotic palpation tactile sensor

Tumor Detection, Segmentation and Classification Challenge on Automated 3D Breast Ultrasound: The TDSC-ABUS Challenge

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Luo, Gongning Xu, Mingwang Chen, Hongyu Liang, Xinjie Tao, Xing Ni, Dong Jeong, Hyunsu Kim, Chulhong Stock, Raphael Baumgartner, Michael Kirchhoff, Yannick Rokuss, Maximilian Maier-Hein, Klaus Yang, Zhikai Fan, Tianyu Boutry, Nicolas Tereshchenko, Dmitry Moine, Arthur Charmetant, Maximilien Sauer, Jan Du, Hao Bai, Xiang-Hui Raikar, Vipul Pai Montoya-Del-Angel, Ricardo Martí, Robert Luna, Miguel Lee, Dongmin Qayyum, Abdul Mazher, Moona Guo, Qihui Wang, Changyan Awasthi, Navchetan Zhao, Qiaochu Wang, Wei Wang, Kuanquan Wang, Qiucheng Dong, Suyu School of Computer Science and Technology Harbin Institute of Technology Harbin150001 China Department of Mathematics Faculty of Science National University of Singapore Singapore National-Regional Key Technology Engineering Laboratory for Medical Ultrasound School of Biomedical Engineering Shenzhen University Medical School Shenzhen University Shenzhen China Laboratory Shenzhen University Shenzhen China School of Biomedical Engineering and Informatics Nanjing Medical University Nanjing China Pohang Korea Republic of Heidelberg Division of Medical Image Computing Heidelberg Germany Faculty of Mathematics and Computer Science Heidelberg University Germany Heidelberg Germany HIDSS4Health - Helmholtz Information and Data Science School for Health Karlsruhe Heidelberg Germany Pattern Analysis and Learning Group Department of Radiation Oncology Heidelberg University Hospital Germany Department of Biomedical Engineering and Health KTH Royal Institute of Technology Stockholm Sweden France FathomX Singapore Saw Swee Hock School of Public Health National University of Singapore Singapore Philips Research University of Girona Spain Department of Robotics and Mechatronics Engineering DGIST Korea Republic of Department of Interdisciplinary Studies of Artificial Intelligence DGIST Korea Republic of National Heart and Lung Institute Faculty of Medicine Imperial College London London United Kingdom Centre for Medical Image Computing Department of Computer Science University College London London United Kingdom Lab School of Communication and Information Engineering Shanghai University Shanghai China Faculty of Science Mathematics and Computer Science Informatics Institute University of Amsterdam Amsterdam1090 GH Netherlands Department of Biomedical Engineering and Physics Amsterdam UMC Amsterdam1081 HV Netherlands Xi’an Jiaotong-Liverpool University China Department of Ultrasound Harbin Medical University Cancer Hospital No. 150 Haping Road Nangang

Breast cancer is one of the most common causes of death among women worldwide. Early detection helps in reducing the number of deaths. Automated 3D Breast Ultrasound (ABUS) is a newer approach for breast screening, which has many advantages over handheld mammography such as safety, speed, and higher detection rate of breast cancer. Tumor detection, segmentation, and classification are key components in the analysis of medical images, especially challenging in the context of 3D ABUS due to the significant variability in tumor size and shape, unclear tumor boundaries, and a low signal-to-noise ratio. The lack of publicly accessible, well-labeled ABUS datasets further hinders the advancement of systems for breast tumor analysis. Addressing this gap, we have organized the inaugural Tumor Detection, Segmentation, and Classification Challenge on Automated 3D Breast Ultrasound 2023 (TDSC-ABUS2023). This initiative aims to spearhead research in this field and create a definitive benchmark for tasks associated with 3D ABUS image analysis. In this paper, we summarize the top-performing algorithms from the challenge and provide critical analysis for ABUS image examination. We offer the TDSC-ABUS challenge as an open-access platform at https://***/ to benchmark and inspire future developments in1algorithmic research. © 2025, CC BY.

关键词： Image segmentation

AdaXpert: Adapting neural architecture for growing data

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Niu, Shuaicheng Wu, Jiaxiang Xu, Guanghui Zhang, Yifan Guo, Yong Zhao, Peilin Wang, Peng Tan, Mingkui School of Software Engineering South China University of Technology China Key Laboratory of Big Data and Intelligent Robot Ministry of Education China Tencent AI Lab China National University of Singapore Singapore Northwestern Polytechnical University China Pazhou Laboratory China

In real-world applications, data often come in a growing manner, where the data volume and the number of classes may increase dynamically. This will bring a critical challenge for learning: given the increasing data volume or the number of classes, one has to instantaneously adjust the neural model capacity to obtain promising performance. Existing methods either ignore the growing nature of data or seek to independently search an optimal architecture for a given dataset, and thus are incapable of promptly adjusting the architectures for the changed data. To address this, we present a neural architecture adaptation method, namely Adaptation eXpert (AdaXpert), to efficiently adjust previous architectures on the growing data. Specifically, we introduce an architecture adjuster to generate a suitable architecture for each data snapshot, based on the previous architecture and the different extent between current and previous data distributions. Furthermore, we propose an adaptation condition to determine the necessity of adjustment, thereby avoiding unnecessary and time-consuming adjustments. Extensive experiments on two growth scenarios (increasing data volume and number of classes) demonstrate the effectiveness of the proposed method. Copyright © 2021, The Authors. All rights reserved.

关键词： Machine learning

Refining BERT embeddings for document hashing via mutual information maximization

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Ou, Zijing Su, Qinliang Yu, Jianxing Zhao, Ruihui Zheng, Yefeng Liu, Bang School of Computer Science and Engineering Sun Yat-sen University Guangzhou China School of Artificial Intelligence Sun Yat-sen University Guangdong China Tencent Jarvis Lab Rali & Mila Université de Montréal Guangdong Key Lab. of Big Data Analysis and Processing Guangzhou China Key Lab. of Machine Intelligence and Advanced Computing Ministry of Education China

Existing unsupervised document hashing methods are mostly established on generative models. Due to the difficulties of capturing long dependency structures, these methods rarely model the raw documents directly, but instead to model the features extracted from them (e.g. bag-of-words (BOW), TFIDF). In this paper, we propose to learn hash codes from BERT embeddings after observing their tremendous successes on downstream tasks. As a first try, we modify existing generative hashing models to accommodate the BERT embeddings. However, little improvement is observed over the codes learned from the old BOW or TFIDF features. We attribute this to the reconstruction requirement in the generative hashing, which will enforce irrelevant information that is abundant in the BERT embeddings also compressed into the codes. To remedy this issue, a new unsupervised hashing paradigm is further proposed based on the mutual information (MI) maximization principle. Specifically, the method first constructs appropriate global and local codes from the documents and then seeks to maximize their mutual information. Experimental results on three benchmark datasets demonstrate that the proposed method is able to generate hash codes that outperform existing ones learned from BOW features by a substantial margin. Copyright © 2021, The Authors. All rights reserved.

关键词： Embeddings

The Minority Matters: A Diversity-Promoting Collaborative Metric Learning Algorithm

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Bao, Shilong Xu, Qianqian Yang, Zhiyong He, Yuan Cao, Xiaochun Huang, Qingming State Key Laboratory of Information Security Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China Key Lab. of Intelligent Information Processing Institute of Computing Technology CAS China School of Computer Science and Tech. University of Chinese Academy of Sciences China Alibaba Group China School of Cyber Science and Technology Shenzhen Campus Sun Yat-sen University China Key Laboratory of Big Data Mining and Knowledge Management CAS China Peng Cheng Laboratory China

Collaborative Metric Learning (CML) has recently emerged as a popular method in recommendation systems (RS), closing the gap between metric learning and Collaborative Filtering. Following the convention of RS, existing methods exploit unique user representation in their model design. This paper focuses on a challenging scenario where a user has multiple categories of interests. Under this setting, we argue that the unique user representation might induce preference bias, especially when the item category distribution is imbalanced. To address this issue, we propose a novel method called Diversity-Promoting Collaborative Metric Learning (DPCML), with the hope of considering the commonly ignored minority interest of the user. The key idea behind DPCML is to include a multiple set of representations for each user in the system. Based on this embedding paradigm, user preference toward an item is aggregated from different embeddings by taking the minimum item-user distance among the user embedding set. Furthermore, we observe that the diversity of the embeddings for the same user also plays an essential role in the model. To this end, we propose a diversity control regularization term to accommodate the multi-vector representation strategy better. Theoretically, we show that DPCML could generalize well to unseen test data by tackling the challenge of the annoying operation that comes from the minimum value. Experiments over a range of benchmark datasets speak to the efficacy of DPCML. © 2022, CC BY-NC-SA.

关键词： Embeddings