检索结果-内蒙古大学图书馆

Improving Continual Relation Extraction through Prototypical Contrastive Learning

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Hu, Chengwei Yang, Deqing Jin, Haoliang Chen, Zhen Xiao, Yanghua School of Data Science Fudan University Shanghai China School of Computer Science Fudan University Shanghai China

Continual relation extraction (CRE) aims to extract relations towards the continuous and iterative arrival of new data, of which the major challenge is the catastrophic forgetting of old tasks. In order to alleviate this critical problem for enhanced CRE performance, we propose a novel Continual Relation Extraction framework with Contrastive Learning, namely CRECL, which is built with a classification network and a prototypical contrastive network to achieve the incremental-class learning of CRE. Specifically, in the contrastive network a given instance is contrasted with the prototype of each candidate relations stored in the memory module. Such contrastive learning scheme ensures the data distributions of all tasks more distinguishable, so as to alleviate the catastrophic forgetting further. Our experiment results not only demonstrate our CRECL’s advantage over the state-of-the-art baselines on two public datasets, but also verify the effectiveness of CRECL’s contrastive learning on improving CRE performance. © 2022, CC BY.

关键词： Extraction

Indoor Navigation Using Ultra Wide Band

学校读者我要写书评

暂无评论

Indoor Navigation Using Ultra Wide Band

Intelligent Computing and Control for Engineering and Business Systems (ICCEBS)

作者： Divya B T Veeramakali S Revathi Harisharajan R R Adhirooban D Ebishdon G V Dept of Computer Science and Engineering Sri Sai Ram Institute of Technology Dept of Data Science and Business Systems School of Computing SRM Institute of Science and Technology Kattankulathur Tamilnadu India

Indoor positioning is a significant and intriguing topic in navigation systems that present numerous use cases for investigation. Researchers are exploring technologies capable of providing accurate locations in indoor spaces where GNSS positioning systems are ineffective. In the numerous solutions available for indoor location services, Ultra-wideband (UWB) technology was chosen for its ability to provide high-precision positioning information for multiple points in real-time. This paper presents a UWB indoor positioning system that utilizes time of flight (TOF) algorithm. At times, time difference of arrival (TDOA) algorithms compute range measurements, this facilitates the trans-receiver (TAG) plane position algorithm's location determination. UWB technology was chosen because it is the latest technology available on Android 12 and further also offers high accuracy, resistance to interference and minimal signal reflection from walls. UltraWide Band(UWB) localization encounters various obstacles, such as multipath distortions, additional signal delays, time discrepancies in clocks, signal interference, and the computational resources necessary to determine the user's location.

关键词：

Imitation Learning from Purified Demonstrations

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wang, Yunke Dong, Minjing Zhao, Yukun Du, Bo Xu, Chang School of Computer Science National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence Wuhan Institute of Data Intelligence Wuhan University China Department of Computer Science City University of Hong Kong China School of Computer Science Faculty of Engineering The University of Sydney Australia

Imitation learning has emerged as a promising approach for addressing sequential decision-making problems, with the assumption that expert demonstrations are optimal. However, in real-world scenarios, most demonstrations are often imperfect, leading to challenges in the effectiveness of imitation learning. While existing research has focused on optimizing with imperfect demonstrations, the training typically requires a certain proportion of optimal demonstrations to guarantee performance. To tackle these problems, we propose to purify the potential noises in imperfect demonstrations first, and subsequently conduct imitation learning from these purified demonstrations. Motivated by the success of diffusion model, we introduce a two-step purification via diffusion process. In the first step, we apply a forward diffusion process to smooth potential noises in imperfect demonstrations by introducing additional noise. Subsequently, a reverse generative process is utilized to recover the optimal demonstration from the diffused ones. We provide theoretical evidence supporting our approach, demonstrating that the distance between the purified and optimal demonstration can be bounded. Empirical results on MuJoCo and RoboSuite demonstrate the effectiveness of our method from different aspects. Copyright © 2023, The Authors. All rights reserved.

关键词： Contrastive Learning

Size-Invariance Matters: Rethinking Metrics and Losses for Imbalanced Multi-object Salient Object Detection 41

学校读者我要写书评

暂无评论

Size-Invariance Matters: Rethinking Metrics and Losses for I...

41st International Conference on Machine Learning, ICML 2024

作者： Li, Feiran Xu, Qianqian Bao, Shilong Yang, Zhiyong Cong, Runmin Cao, Xiaochun Huang, Qingming Institute of Information Engineering Chinese Academy of Sciences Beijing China School of Cyber Security University of Chinese Academy of Sciences Beijing China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing China School of Computer Science and Technology University of Chinese Academy of Sciences Beijing China Institute of Information Science Beijing Jiaotong University Beijing China School of Control Science and Engineering Shandong University Jinan China Key Laboratory of Machine Intelligence and System Control Ministry of Education Jinan China School of Cyber Science and Tech. Sun Yat-sen University Shenzhen Campus China Key Laboratory of Big Data Mining and Knowledge Management Chinese Academy of Sciences Beijing China

This paper explores the size-invariance of evaluation metrics in Salient Object Detection (SOD), especially when multiple targets of diverse sizes co-exist in the same image. We observe that current metrics are size-sensitive, where larger objects are focused, and smaller ones tend to be ignored. We argue that the evaluation should be size-invariant because bias based on size is unjustified without additional semantic information. In pursuit of this, we propose a generic approach that evaluates each salient object separately and then combines the results, effectively alleviating the imbalance. We further develop an optimization framework tailored to this goal, achieving considerable improvements in detecting objects of different sizes. Theoretically, we provide evidence supporting the validity of our new metrics and present the generalization analysis of SOD. Extensive experiments demonstrate the effectiveness of our method. The code is available at https://***/Ferry-Li/SI-SOD. Copyright 2024 by the author(s)

关键词： Object detection

Classification of Alzheimer’s Disease using Random Oversampling and Albumentations on Convolutional Neural Network

学校读者我要写书评

暂无评论

Classification of Alzheimer’s Disease using Random Oversamp...

International Conference on Informatics and Computing (ICIC)

作者： Calvin Kamardi I Kadek Perry Bagus Laksana Maria Susan Anggreainy Nora Fitriawati Michiko Amaya Yonatan Tesalonika Abigail Eikwine Mangkang Vianny Pengesa Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia Computer Science Department BINUS Graduate Program Computer Science Bina Nusantara University Jakarta Indonesia Information Systems Department BINUS Online Learning Bina Nusantara University Jakarta Indonesia

Alzheimer’s Disease (AD) is a brain disorder that causes dementia and affects the memory, cognitive, and behavioral function. Early detection for AD can help to reduce the symptoms and slow down AD progression. Deep learning, particularly Convolutional Neural Network (CNN), a component of artificial intelligence, has demonstrated efficacy in resolving image-related issues and has gained widespread use in the analysis of medical images. CNN works well on large datasets, however, in imbalanced datasets, it can perform poorly by misclassifying minority classes. In this study, the random over-sampling technique is used to address the imbalance in the dataset, while Albumentations is used for image augmentation to prevent overfitting. The results demonstrate that the combination of these techniques leads to improved classification performance compared to traditional approaches. The proposed method has the potential to improve the accuracy of AD diagnosis and contribute to the development of effective treatments for this debilitating condition.

关键词：

Unveiling the Truth and Facilitating Change: Towards Agent-based Large-scale Social Movement Simulation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Mou, Xinyi Wei, Zhongyu Huang, Xuanjing School of Data Science Fudan University China Research Institute of Intelligent and Complex Systems Fudan University China School of Computer Science Fudan University China Shanghai Collaborative Innovation Center of Intelligent Visual Computing China

Social media has emerged as a cornerstone of social movements, wielding significant influence in driving societal change. Simulating the response of the public and forecasting the potential impact has become increasingly important. However, existing methods for simulating such phenomena encounter challenges concerning their efficacy and efficiency in capturing the behaviors of social movement participants. In this paper, we introduce a hybrid framework HiSim for social media user simulation, wherein users are categorized into two types. Core users are driven by Large Language Models, while numerous ordinary users are modeled by deductive agent-based models. We further construct a Twitter-like environment to replicate their response dynamics following trigger events. Subsequently, we develop a multi-faceted benchmark SoMoSiMu-Bench for evaluation and conduct comprehensive experiments across real-world datasets. Experimental results demonstrate the effectiveness and flexibility of our method 1 Copyright © 2024, The Authors. All rights reserved.

关键词： Computational methods

A multip-level additive distortion method for security improvement in palette image steganography

学校读者我要写书评

暂无评论

Journal of Visual Communication and Image Representation 2025年 110卷

作者： Chen, Yi Wang, Hongxia Cui, Yunhe Shen, Guowei Guo, Chun Liu, Yong Wu, Hanzhou State Key Laboratory of Public Big Data College of Computer Science and Technology Guizhou University Guiyang550025 China Guizhou Provincial Characteristic Key Laboratory of Software Engineering and Information Security Guizhou University Guiyang550025 China School of Cyber Science and Engineering Sichuan University Chengdu610065 China School of Artificial Intelligence Hubei University of Automotive Technology Shiyan442002 China School of Big Data and Computer Science Guizhou Normal University Guiyang550025 China School of Communication and Information Engineering Shanghai University Shanghai200444 China

With the rapid development of the Internet and communication technology, palette images have become a preferred media for steganography. However, the security of palette image steganography faces a big problem. To address this, we propose a multiple-level additive distortion method for security improvement in palette image steganography. The proposed multiple-level additive distortion method comprises an index-level cost method and a pixel-level cost method. The index-level and the pixel-level costs by the two methods can respectively reflect the relationship changes of adjacent indices and the pixels corresponding to the adjacent indices. Meanwhile, the index-level and the pixel-level costs can also reflect the modification impact of steganography. Therefore, the proposed method can improve the security of palette image steganography. We conducted extensive experiments on three datasets to verify the security improvement. Experiment results have shown our proposed multiple-level distortion method indeed has an advantage in security when compared with four state-of-the-art methods. © 2025 Elsevier Inc.

关键词： Additive costs Color table Index matrix Multiple-level distortions Palette images Security Steganography

Deep learning model for pleural effusion detection via active learning and pseudo-labeling: a multisite study

学校读者我要写书评

暂无评论

BMC Medical Imaging 2024年第1期24卷 92页

作者： Chang, Joseph Lin, Bo-Ru Wang, Ti-Hao Chen, Chung-Ming Department of Biomedical Engineering College of Medicine and College of Engineering National Taiwan University No. 1 Sec. 1 Jen-Ai Road Taipei 100 Taipei 100 Taiwan EverFortune.AI Co. Ltd Taichung Taiwan The Data Science Degree Program College of Electrical Engineering and Computer Science National Taiwan University and Academia Sinica Taipei Taiwan Department of Radiation Oncology China Medical University Hospital Taichung Taiwan Department of Medicine China Medical University Taichung Taiwan

Background: The study aimed to develop and validate a deep learning-based computer Aided Triage (CADt) algorithm for detecting pleural effusion in chest radiographs using an active learning (AL) framework. This is aimed at addressing the critical need for a clinical grade algorithm that can timely diagnose pleural effusion, which affects approximately 1.5 million people annually in the United States. Methods: In this multisite study, 10,599 chest radiographs from 2006 to 2018 were retrospectively collected from an institution in Taiwan to train the deep learning algorithm. The AL framework utilized significantly reduced the need for expert annotations. For external validation, the algorithm was tested on a multisite dataset of 600 chest radiographs from 22 clinical sites in the United States and Taiwan, which were annotated by three U.S. board-certified radiologists. Results: The CADt algorithm demonstrated high effectiveness in identifying pleural effusion, achieving a sensitivity of 0.95 (95% CI: [0.92, 0.97]) and a specificity of 0.97 (95% CI: [0.95, 0.99]). The area under the receiver operating characteristic curve (AUC) was 0.97 (95% DeLong’s CI: [0.95, 0.99]). Subgroup analyses showed that the algorithm maintained robust performance across various demographics and clinical settings. Conclusion: This study presents a novel approach in developing clinical grade CADt solutions for the diagnosis of pleural effusion. The AL-based CADt algorithm not only achieved high accuracy in detecting pleural effusion but also significantly reduced the workload required for clinical experts in annotating medical data. This method enhances the feasibility of employing advanced technological solutions for prompt and accurate diagnosis in medical settings. © The Author(s) 2024.

关键词： Active learning Chest radiographs Deep learning Pleural effusion X-rays

Question-Answer Cross Language Image Matching for Weakly Supervised Semantic Segmentation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Deng, Songhe Zhuo, Wei Xie, Jinheng Shen, Linlin Computer Vision Institute School of Computer Science and Software Engineering Shenzhen University China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China

Class Activation Map (CAM) has emerged as a popular tool for weakly supervised semantic segmentation (WSSS), allowing the localization of object regions in an image using only image-level labels. However, existing CAM methods suffer from under-activation of target object regions and false-activation of background regions due to the fact that a lack of detailed supervision can hinder the model’s ability to understand the image as a whole. In this paper, we propose a novel Question-Answer Cross-Language-Image Matching framework for WSSS (QA-CLIMS), leveraging the vision-language foundation model to maximize the text-based understanding of images and guide the generation of activation maps. First, a series of carefully designed questions are posed to the VQA (Visual Question Answering) model with Question-Answer Prompt Engineering (QAPE) to generate a corpus of both foreground target objects and backgrounds that are adaptive to query images. We then employ contrastive learning in a Region Image Text Contrastive (RITC) network to compare the obtained foreground and background regions with the generated corpus. Our approach exploits the rich textual information from the open vocabulary as additional supervision, enabling the model to generate high-quality CAMs with a more complete object region and reduce false-activation of background regions. We conduct extensive analysis to validate the proposed method and show that our approach performs state-of-the-art on both PASCAL VOC 2012 and MS COCO datasets. Code is available at: https://***/CVI-SZU/QA-CLIMS Copyright © 2024, The Authors. All rights reserved.

关键词： Chemical activation