检索结果-内蒙古大学图书馆

No One Left Behind: Real-World Federated Class-Incremental Learning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Dong, Jiahua Li, Hongliu Cong, Yang Sun, Gan Zhang, Yulun Van Gool, Luc The State Key Laboratory of Robotics Shenyang Institute of Automation Chinese Academy of Sciences Shenyang110016 China The Institutes for Robotics and Intelligent Manufacturing Chinese Academy of Sciences Shenyang110169 China The University of Chinese Academy of Sciences Beijing100049 China The Department of Civil and Environmental Engineering Hong Kong Polytechnic University Hong Kong The College of Automation Science and Engineering South China University of Technology Guangzhou510640 China The Computer Vision Lab ETH Zürich Zürich8092 Switzerland

Federated learning (FL) is a hot collaborative training framework via aggregating model parameters of decentralized local clients. However, most FL methods unreasonably assume data categories of FL framework are known and fixed in advance. Moreover, some new local clients that collect novel categories unseen by other clients may be introduced to FL training irregularly. These issues render global model to undergo catastrophic forgetting on old categories, when local clients receive new categories consecutively under limited memory of storing old categories. To tackle the above issues, we propose a novel Local-Global Anti-forgetting (LGA) model. It ensures no local clients are left behind as they learn new classes continually, by addressing local and global catastrophic forgetting. Specifically, considering tackling class imbalance of local client to surmount local forgetting, we develop a category-balanced gradient-adaptive compensation loss and a category gradient-induced semantic distillation loss. They can balance heterogeneous forgetting speeds of hard-to-forget and easy-to-forget old categories, while ensure consistent class-relations within different tasks. Moreover, a proxy server is designed to tackle global forgetting caused by Non-IID class imbalance between different clients. It augments perturbed prototype images of new categories collected from local clients via self-supervised prototype augmentation, thus improving robustness to choose the best old global model for local-side semantic distillation loss. Experiments on representative datasets verify superior performance of our model against comparison methods. The code is available at https://***/JiahuaDong/LGA. Copyright © 2023, The Authors. All rights reserved.

关键词： Distillation

Neural Gradient Regularizer

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Xu, Shuang Wang, Yifan Zhao, Zixiang Peng, Jiangjun Cao, Xiangyong Meng, Deyu Zhang, Yulun Timofte, Radu Van Gool, Luc School of Mathematics and Statistics Northwestern Polytechnical University Xi’an710021 China Research and Development Institute Northwestern Polytechnical University in Shenzhen Shenzhen518063 China School of Electronic and Information Engineering The Key Laboratory for Intelligent Networks and Network Security Ministry of Education Xi’an Jiaotong University Xi’an710049 China School of Mathematics and Statistics Xi’an Jiaotong University Xi’an710049 China Computer Vision Lab ETH Zurich Zürich8092 Switzerland

Owing to its significant success, the prior imposed on gradient maps has consistently been a subject of great interest in the field of image processing. Total variation (TV), one of the most representative regularizers, is known for its ability to capture the intrinsic sparsity prior underlying gradient maps. Nonetheless, TV and its variants often underestimate the gradient maps, leading to the weakening of edges and details whose gradients should not be zero in the original image (i.e., image structures is not describable by sparse priors of gradient maps). Recently, total deep variation (TDV) has been introduced, assuming the sparsity of feature maps, which provides a flexible regularization learned from large-scale datasets for a specific task. However, TDV requires to retrain the network with image/task variations, limiting its versatility. To alleviate this issue, in this paper, we propose a neural gradient regularizer (NGR) that expresses the gradient map as the output of a neural network. Unlike existing methods, NGR does not rely on any subjective sparsity or other prior assumptions on image gradient maps, thereby avoiding the underestimation of gradient maps. NGR is applicable to various image types and different image processing tasks, functioning in a zero-shot learning fashion, making it a versatile and plug-and-play regularizer. Extensive experimental results demonstrate the superior performance of NGR over state-of-the-art counterparts for a range of different tasks, further validating its effectiveness and versatility. © 2023, CC BY.

关键词： Large dataset

HIPA: Hierarchical Patch Transformer for Single Image Super Resolution

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Cai, Qing Qian, Yiming Li, Jinxing Lyu, Jun Yang, Yee-Hong Wu, Feng Zhang, David The Faculty of Information Science and Engineering Ocean University of China Shandong Qingdao266100 China The Department of Computer Science University of Manitoba WinnipegMBR3T 2N2 Canada The School of Computer Science and Technology Harbin Institute of Technology Guangdong Shenzhen518055 China The School of Nursing The Hong Kong Polytechnic University Hong Kong The Department of Computing Science University of Alberta EdmontonABT6G 2E9 Canada The School of Information Science and Technology University of Science and Technology of China Anhui Hefei230026 China The School of Data Science The Chinese University of Hong Kong Guangdong Shenzhen518172 China The Shenzhen Institute of Artificial Intelligence and Robotics for Society Guangdong Shenzhen518000 China Linklogis Joint Laboratory of Computer Vision and Artificial Intelligence Guangdong Shenzhen518172 China

Transformer-based architectures start to emerge in single image super resolution (SISR) and have achieved promising performance. However, most existing vision Transformer-based SISR methods still have two shortcomings: (1) they divide images into the same number of patches with a fixed size, which may not be optimal for restoring patches with different levels of texture richness;and (2) their position encodings treat all input tokens equally and hence, neglect the dependencies among them. This paper presents a HIPA, which stands for a novel Transformer architecture that progressively recovers the high resolution image using a hierarchical patch partition. Specifically, we build a cascaded model that processes an input image in multiple stages, where we start with tokens with small patch sizes and gradually merge them to form the full resolution. Such a hierarchical patch mechanism not only explicitly enables feature aggregation at multiple resolutions but also adaptively learns patch-aware features for different image regions, e.g., using a smaller patch for areas with fine details and a larger patch for textureless regions. Meanwhile, a new attention-based position encoding scheme for Transformer is proposed to let the network focus on which tokens should be paid more attention by assigning different weights to different tokens, which is the first time to our best knowledge. Furthermore, we also propose a multi-receptive field attention module to enlarge the convolution receptive field from different branches. The experimental results on several public datasets demonstrate the superior performance of the proposed HIPA over previous methods quantitatively and qualitatively. We will share our code and models when the paper is accepted. Copyright © 2022, The Authors. All rights reserved.

关键词： Image reconstruction

Perceptive self-supervised learning network for noisy image watermark removal

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Tian, Chunwei Zheng, Menghua Li, Bo Zhang, Yanning Zhang, Shichao Zhang, David School of Software Northwestern Polytechnical University Xi’an710129 China National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology Xi’an710129 China School of Electronics and Information Northwestern Polytechnical University Xi’an710129 China School of Computer Science Northwestern Polytechnical University National Engineering Laboratory for Integrated AeroSpace-Ground-Ocean Big Data Application Technology Xi’an710129 China Guangxi Key Lab of Multisource Information Mining & Security College of Computer Science & Engineering Guangxi Normal University Guilin541004 China Shenzhen518172 China Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen518172 China

Popular methods usually use a degradation model in a supervised way to learn a watermark removal model. However, it is true that reference images are difficult to obtain in the real world, as well as collected images by cameras suffer from noise. To overcome these drawbacks, we propose a perceptive self-supervised learning network for noisy image watermark removal (PSLNet) in this paper. PSLNet depends on a parallel network to remove noise and watermarks. The upper network uses task decomposition ideas to remove noise and watermarks in sequence. The lower network utilizes the degradation model idea to simultaneously remove noise and watermarks. Specifically, mentioned paired watermark images are obtained in a self-supervised way, and paired noisy images (i.e., noisy and reference images) are obtained in a supervised way. To enhance the clarity of obtained images, interacting two sub-networks and fusing obtained clean images are used to improve the effects of image watermark removal in terms of structural information and pixel enhancement. Taking into texture information account, a mixed loss uses obtained images and features to achieve a robust model of noisy image watermark removal. Comprehensive experiments show that our proposed method is very effective in comparison with popular convolutional neural networks (CNNs) for noisy image watermark removal. Codes can be obtained at https://***/hellloxiaotian/PSLNet. Copyright © 2024, The Authors. All rights reserved.

关键词： Image denoising

Efficient Multi-Query Oriented Continuous Subgraph Matching

学校读者我要写书评

暂无评论

Efficient Multi-Query Oriented Continuous Subgraph Matching

International Conference on Data Engineering

作者： Ziyi Ma Jianye Yang Xu Zhou Guoqing Xiao Jianhua Wang Liang Yang Kenli Li Xuemin Lin School of Artificial Intelligence Hebei University of Technology China Guangxi Key Laboratory of Machine Vision and Intelligent Control Wuzhou University China Cyberspace Institute of Advanced Technology Guangzhou University China Department of New Networks PengCheng Laboratory China College of Computer Science and Electronic Engineering Hunan University China Shenzhen Research Institute Hunan University China Antai College of Economics and Management Shanghai Jiao Tong University China

ISBN: (数字)9798350317152

ISBN: (纸本)9798350317169

Continuous subgraph matching (CSM) is a critical task for analyzing dynamic graphs and has a wide range of applications, such as merchant fraud detection, cyber-attack hunting, and rumor detection. Although many efficient CSM algorithms have been recently proposed, they are mainly designed to process a single query. However, in some application scenarios, multi-query oriented continuous subgraph matching (MQCSM) may be of more interest. To our knowledge, the two existing solutions to MQCSM are outdated due to unsatisfactory performance. In this paper, we propose MQ-Match, an efficient approach to MQCSM. First, we design a compact yet effective index structure CCG, which maintains the local matching result of vertices in the data graph using a directed graph. The directed edges in CCG can be utilized as an effective pruning rule for the subsequent incremental matching algorithm when expanding a partial match. Then, we develop a computation sharing incremental matching algorithm. In specific, a set of matching trees is constructed based on the depth-first search trees of the query graphs. By utilizing CCG, we conduct subgraph matching for the matching tree to collect the incremental matches for the query graphs, where the common structures of query graphs are matched only once. Extensive experiments show that MQ-Match can achieve 3.1x-7071.4x speedup over the competitors, and consumes much less memory under the majority of the experiment settings.

关键词： Heuristic algorithms Memory management Directed graphs Data engineering Fraud Computational efficiency Indexes

Tactile-Based Object Pose Estimation Employing Extended Kalman Filter

学校读者我要写书评

暂无评论

Tactile-Based Object Pose Estimation Employing Extended Kalm...

International Conference on Advanced robotics and Mechatronics (ICARM)

作者： Qiguang Lin Chaojie Yan Qiang Li Yonggen Ling Yu Zheng Wangwei Lee Zhaoliang Wan Bidan Huang Xiaofeng Liu Jiangsu Key laboratory of Special Robotic Technology College of IoT Engineering Hohai University Changzhou Jiangsu P.R. China State Key Laboratory of Industrial Control and Technology Zhejiang University Institute of Cyber-System and Control Hangzhou P.R. China Neuroinformatics Group Center for Cognitive Interaction Technology (CITEC) Bielefeld University Bielefeld Germany Tencent Robotics X Shenzhen China School of Computer Science and Engineering Sun Yat-sen University Guangzhou P.R.China

In this paper, we present a new approach to estimate the pose of an object being manipulated by a multi-fingered robotic hand. The method utilizes advanced tactile sensors with high spatial resolution to optimize the estimation of the object's pose using an Extended Kalman Filter (EKF) based approach. We defined and derived the state and measurement equations, as well as evaluated the estimation accuracy in grasping tasks. The approach is able to effectively account for the pose transition caused by tactile pushing, and the mapping from the object's pose to the contact position and normal direction as measured by the tactile sensor. The method was evaluated in multiple grasping experiments in simulation scenarios. Results show that the estimation can converge towards the ground truth in a relatively short period of time, with displacement and rotation errors remaining within acceptable levels. This new method has the potential to improve the accuracy and reliability of robotic grasping and manipulation tasks.

关键词：

Long and Short-Term Constraints Driven Safe Reinforcement Learning for Autonomous Driving

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hu, Xuemin Chen, Pan Wen, Yijun Tang, Bo Chen, Long School of Artificial Intelligence Hubei University Wuhan430062 China Key Laboratory of Intelligent Sensing System and Security Hubei University Ministry of Education Hubei Wuhan430062 China Department of Electrical and Computer Engineering Worcester Polytechnic Institute WorcesterMA01609 United States State Key Laboratory of Multimodal Artificial Intelligence Systems State Key Laboratory of Management and Control for Complex Systems Chinese Academy of Sciences Beijing100190 China WAYTOUS Inc. Beijing100083 China Shenzhen518107 China Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University Xi’an710049 China

Reinforcement learning (RL) has been widely used in decision-making and control tasks, but the risk is very high for the agent in the training process due to the requirements of interaction with the environment, which seriously limits its industrial applications such as autonomous driving systems. Safe RL methods are developed to handle this issue by constraining the expected safety violation costs as a training objective, but the occurring probability of an unsafe state is still high, which is unacceptable in autonomous driving tasks. Moreover, these methods are difficult to achieve a balance between the cost and return expectations, which leads to learning performance degradation for the algorithms. In this paper, we propose a novel algorithm based on the long and short-term constraints (LSTC) for safe RL. The short-term constraint aims to enhance the shortterm state safety that the vehicle explores, while the long-term constraint enhances the overall safety of the vehicle throughout the decision-making process, both of which are jointly used to enhance the vehicle safety in the training process. In addition, we develop a safe RL method with dual-constraint optimization based on the Lagrange multiplier to optimize the training process for end-to-end autonomous driving. Comprehensive experiments were conducted on the MetaDrive simulator. Experimental results demonstrate that the proposed method achieves higher safety in continuous state and action tasks, and exhibits higher exploration performance in long-distance decision-making tasks compared with state-of-the-art methods. Copyright © 2024, The Authors. All rights reserved.

关键词： Lagrange multipliers

V2-SfMLearner: Learning Monocular Depth and Ego-motion for Multimodal Wireless Capsule Endoscopy

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Bai, Long Cui, Beilei Wang, Liangyu Li, Yanheng Yao, Shilong Yuan, Sishen Wu, Yanan Zhang, Yang Meng, Max Q.-H. Li, Zhen Ding, Weiping Ren, Hongliang Dept. of Electronic Engineering The Chinese University of Hong Kong Hong Kong City University of Hong Kong Hong Kong Shenzhen Key Laboratory of Robotics Perception and Intelligence Dept. of Electronic and Electrical Engineering Southern University of Science and Technology Shenzhen China The School of Mechanical Engineering Hubei University of Technology Wuhan China The Dept. of Gastroenterology Qilu Hospital of Shandong University Jinan China The School of Artificial Intelligence and Computer Science Nantong University Nantong China

Deep learning can predict depth maps and capsule ego-motion from capsule endoscopy videos, aiding in 3D scene reconstruction and lesion localization. However, the collisions of the capsule endoscopies within the gastrointestinal tract cause vibration perturbations in the training data. Existing solutions focus solely on vision-based processing, neglecting other auxiliary signals like vibrations that could reduce noise and improve performance. Therefore, we propose V2-SfMLearner, a multimodal approach integrating vibration signals into vision-based depth and capsule motion estimation for monocular capsule endoscopy. We construct a multimodal capsule endoscopy dataset containing vibration and visual signals, and our artificial intelligence solution develops an unsupervised method using vision-vibration signals, effectively eliminating vibration perturbations through multimodal learning. Specifically, we carefully design a vibration network branch and a Fourier fusion module, to detect and mitigate vibration noises. The fusion framework is compatible with popular vision-only algorithms. Extensive validation on the multimodal dataset demonstrates superior performance and robustness against vision-only algorithms. Without the need for large external equipment, our V2-SfMLearner has the potential for integration into clinical capsule robots, providing real-time and dependable digestive examination tools. The findings show promise for practical implementation in clinical settings, enhancing the diagnostic capabilities of doctors. Copyright © 2024, The Authors. All rights reserved.

关键词： Endoscopy

A Prediction Approach Based on Long Short-Term Memory Networks for Dynamic Multiobjective Optimization

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Xu, Biao Rang, Gejie Li, Wenji Gong, Dunwei Fan, Zhun Yang, Shengxiang He, Jie College of Engineering Shantou University Shantou515063 China College of Automation and Electronic Engineering Qingdao University of Science and Technology Qingdao266100 China Guangxi Key Laboratory of Machine Vision and Intelligent Control Wuzhou University Wuzhou543002 China Shenzhen institute for Advanced Study UESTC Shenzhen518110 China School of Computer Science and Informatics De Montfort University LeicesterLE1 9BH United Kingdom Minjiang University China

Dynamic multiobjective optimization problems (DMOPs) present significant challenges to conventional evolutionary optimization methods because of the continuous changes in their Pareto-optimal sets (PSs) and fronts (PFs). Prediction-driven approaches have demonstrated potential in rapidly adapting to these changes. However, many existing methods depend on linear models to forecast the evolving PSs, which may be restrictive. To counteract this limitation, this research presents a novel dynamic multiobjective evolutionary optimization algorithm that incorporates predictions from long short-term memory (LSTM) networks. Initially, in our methodology, the PS for each problem is segmented into multiple clusters, and the centroid of each cluster is identified. These cluster centroids, representing the PSs across various environmental conditions, are then transformed into a series of time series data. The LSTM network models are subsequently trained on this time series data as input samples. Utilizing these refined models, the centroids of the evolving PSs are predicted with improved precision. Moreover, to enhance the performance of the algorithm, an innovative population-generation strategy is also introduced that guarantees a well-converged and diverse starting population. Our proposed algorithm undergoes rigorous testing using benchmark functions, and the outcomes validate its proficiency in tackling DMOPs, showing superior performance compared to existing state-of-the-art algorithms. © 2024, The Authors. All rights reserved.

关键词： Forecasting