检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Wu, Gaojie Zheng, Wei-Shi Lu, Yutong Tian, Qi The School of Computer Science and Engineering Sun Yat-Sen University Guangzhou510275 China The School of Data and Computer Science Sun Yat-Sen University Guangzhou510275 China Peng Cheng Laboratory Shenzhen518005 China The Key Laboratory of Machine Intelligence and Advanced Computing Sun Yat-Sen University Ministry of Education China The Cloud & AI BU Huawei China

Vision Transformer (ViT) has shown great potential for various visual tasks due to its ability to model long-range dependency. However, ViT requires a large amount of computing resource to compute the global self-attention. In this work, we propose a ladder self-attention block with multiple branches and a progressive shift mechanism to develop a light-weight transformer backbone that requires less computing resources (e.g. a relatively small number of parameters and FLOPs), termed Progressive Shift Ladder Transformer (PSLT). First, the ladder self-attention block reduces the computational cost by modelling local self-attention in each branch. In the meanwhile, the progressive shift mechanism is proposed to enlarge the receptive field in the ladder self-attention block by modelling diverse local self-attention for each branch and interacting among these branches. Second, the input feature of the ladder self-attention block is split equally along the channel dimension for each branch, which considerably reduces the computational cost in the ladder self-attention block (with nearly 13 the amount of parameters and FLOPs), and the outputs of these branches are then collaborated by a pixel-adaptive fusion. Therefore, the ladder self-attention block with a relatively small number of parameters and FLOPs is capable of modelling long-range interactions. Based on the ladder self-attention block, PSLT performs well on several vision tasks, including image classification, objection detection and person re-identification. On the ImageNet-1k dataset, PSLT achieves a top-1 accuracy of 79.9% with 9.2M parameters and 1.9G FLOPs, which is comparable to several existing models with more than 20M parameters and 4G FLOPs. Code is available at https://***/wugaojie/***. Copyright © 2023, The Authors. All rights reserved.

关键词： Ladders

来源：评论

学校读者我要写书评

暂无评论

An One-Stage Object Detection Network Based on Image and Feature Enhancement for Automated Defect Inspection

SSRN

引用

SSRN 2023年

作者： Ying, Xiang Hu, Yifan Gao, Jie Fu, Xuzhou Yu, Mei Li, Xuewei College of Intelligence and Computing Tianjin University 135 Yaguan Road Jinnan District Tianjin Tianjin300354 China Tianjin Key Laboratory of Cognitive Computing and Application Tianjin University 135 Yaguan Road Jinnan District Tianjin Tianjin300354 China Tianjin Key Laboratory of Advanced Networking Tianjin University 135 Yaguan Road Jinnan District Tianjin Tianjin300354 China

In recent years, the industrial defect detection method based on deep learning has been widely applied in various scenarios and has achieved remarkable results. However, the imaging quality of industrial images is usually poor, which limits the further improvement of detection performance. Moreover, the insufficient data and the lack of semantics of the defect object make it difficult to capture correct visual features, resulting in error responses in feature maps. To solve the above problems, we propose an one-stage network for defect detection, which contains two supervised attention modules: image enhancement subnet and feature enhancement subnet. By mining the underutilized prior in the annotations, these two subnets are able to improve the image quality and suppress the error responses. Experiments show that the performance of our model is significantly improved compared with that of the baseline, and it also surpasses other representative methods, which proves the effectiveness of our model. © 2023, The Authors. All rights reserved.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Variable Rate Syndrome-Trellis Codes for Steganography on Bursty Channels 19th

Variable Rate Syndrome-Trellis Codes for Steganography on Bu...

引用

19th International Workshop on Digital Forensics and Watermarking, IWDW 2020

作者： Feng, Bingwen Liu, Zhiquan Wei, Kaimin Lu, Wei Lin, Yuchun College of Information Science and Technology Jinan University Guangzhou510632 China State Key Laboratory of Information Security Institute of Information Engineering Chinese Academy of Sciences Beijing100093 China Guangdong Key Laboratory of Intelligent Information Processing and Shenzhen Key Laboratory of Media Security Shenzhen518060 China School of Computer Science and Engineering Guangdong Province Key Laboratory of Information Security Technology Ministry of Education Key Laboratory of Machine Intelligence and Advanced Computing Sun Yat-sen University Guangzhou510006 China Guangzhou Institute of Science and Technology Guangzhou510006 China

ISBN: (纸本)9783030694487

This paper presents a type of variable rate syndrome-trellis codes (VR-STC) for bursty channels. It can embed message bits with two different embedding rates. In the embedding, a cover vector is sliced into segments, and the embedding rates for each segment depend on the local channel distribution. The parities of stego segments are exploited to indicate the selected embedding rates according to a mapping function between parities and embedding rates. The core of the VR-STC is the parity-aware encoder, which can simultaneously output two candidate stego segments with different embedding rates and opposite parities, either of which can be used to constitute the final stego vector. A Viterbi algorithm is also suggested to find the closed stego segments. Besides, the mapping function between parities and embedding rates is designed by minimizing the embedding costs on a down sampled version of the cover vector. It can further improve the undetectability of the VR-STC. Experimental results on artificial signals and binary images suggest that the proposed VR-STC can provide high success rate of embedding and reduce the embedding cost on bursty channels. © 2021, Springer Nature Switzerland AG.

关键词： Steganography

来源：评论

学校读者我要写书评

暂无评论

Skeleton-Embedded Network for action recognition

Skeleton-Embedded Network for action recognition

引用

IEEE International Conference on Real-time computing and Robotics (RCAR)

作者： Li Luo Ziliang Ren Yong Qin Qieshi Zhang Xiangyang Gao Department of Computer Science Dongguan University of Technology Dongguan China CAS Key Laboratory of Human-Machine Intelligence-Synergy Systems Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Shenzhen China The Chinese University of Hong Kong China

Multimodality-based human action recognition is becoming an increasingly attractive topic as different modalities can provide complementary information. RGB and skeleton data have their pros and limitations for action recognition because they describe the action from distinctive views. Therefore, the fusion of these two modalities using their complementarity is meaningful to action recognition. However, it is difficult to completely utilize the supplementarity of RGB and skeleton data. In this paper, to take advantage of their complementary at the semantic level, A skeleton-embedded network(SE-Net) is proposed by us. The architecture is based on integrating their feature in the network not only a post-hoc fusion. With a sparse sampling strategy, which is efficient for learning using the whole action video. To examine the performance of our method, we conducted experiments on the wildly used datasets, obtaining an accuracy of 86.4% in C-sub and 88.2% in C-view on NTU-RGB+D 60. For NTU-RGB+D 120, the accuracy of 81.3% in C-sub and 83.4% in C-set is obtained.

关键词：

来源：评论

学校读者我要写书评

暂无评论

MIMO-DoAnet: Multi-channel Input and Multiple Outputs DoA Network with Unknown Number of Sound Sources

arXiv

引用

arXiv 2022年

作者： Yin, Haoran Ge, Meng Fu, Yanjie Zhang, Gaoyan Wang, Longbiao Zhang, Lei Qiu, Lin Dang, Jianwu Tianjin Key Laboratory of Cognitive Computing and Application College of Intelligence and Computing Tianjin University Tianjin China Ict Products & Solutions Huawei Dongguan China Japan Advanced Institute of Science and Technology Ishikawa Japan

Recent neural network based Direction of Arrival (DoA) estimation algorithms have performed well on unknown number of sound sources scenarios. These algorithms are usually achieved by mapping the multi-channel audio input to the single output (i.e. overall spatial pseudo-spectrum (SPS) of all sources), that is called MISO. However, such MISO algorithms strongly depend on empirical threshold setting and the angle assumption that the angles between the sound sources are greater than a fixed angle. To address these limitations, we propose a novel multi-channel input and multiple outputs DoA network called MIMO-DoAnet. Unlike the general MISO algorithms, MIMODoAnet predicts the SPS coding of each sound source with the help of the informative spatial covariance matrix. By doing so, the threshold task of detecting the number of sound sources becomes an easier task of detecting whether there is a sound source in each output, and the serious interaction between sound sources disappears during inference stage. Experimental results show that MIMO-DoAnet achieves relative 18.6% and absolute 13.3%, relative 34.4% and absolute 20.2% F1 score improvement compared with the MISO baseline system in 3, 4 sources scenes. The results also demonstrate MIMO-DoAnet alleviates the threshold setting problem and solves the angle assumption problem effectively. Copyright © 2022, The Authors. All rights reserved.

关键词： Direction of arrival

来源：评论

学校读者我要写书评

暂无评论

A Preliminary Study on the Functional Coupling between Nerve and Blood Microcirculation for Applications in Rehabilitation Robots

A Preliminary Study on the Functional Coupling between Nerve...

引用

IEEE International Conference on Robotics and Biomimetics

作者： Qingge Li Yuanzhe Dong Yuxiang Zhang Xin Wang Naifu Jiang Jianping Huang Han Cui Lan Tian Yue Zheng Xiangxin Li Lin Wang Guanglin Li Wenyuan Liang Liang Peng Peng Fang CAS Key Laboratory of Human-Machine Intelligence-Synergy Systems Shenzhen Institute of Advanced Technology Shenzhen China Shenzhen College of Advanced Technology University of Chinese Academy of Sciences Shenzhen China National Research Center for Rehabilitation Technical Aids Beijing China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China

Rehabilitation robots play an important role in the motor function rehabilitation for stroke survivors with hemiplegia. However, the rehabilitation effect of current robots is still limited partly because a single training of motor function can be strongly affected by the decreased blood supply function of the bedridden patients. This work proposed an approach to study the coupling relationship between the motor and blood supply functions by combining the synchronously recorded EEG and cerebral blood oxygen information, where the cerebrations in different movement paradigms were analyzed from an aspect of "functional coupling". The results show that the information of oxyhemoglobin concentration change (ΔHbO) can effectively indicate the cortex activation, and a stronger blood supply is needed in cortexes to perform body movements. The correlations within motor cortexes are significantly stronger than the ones between motor and prefrontal cortexes, and a higher resistance level of extremity training will cause stronger correlations. Calculation of transfer entropy (TE) shows that there exists a bidirectional information transmission between the electrophysiological and blood supply signals, and more information is transmitted always in the direction from ΔHbO to EEG than in the opposite direction. The information transmission or the coupling relationship can be significantly enhanced by extremity movements, and large TE values are achieved in the Theta, Beta and Gamma frequency bands of EEG that correspond to motor functions. This work has demonstrated the functional coupling between nerve and blood microcirculation, which would provide a technical guidance to improve the rehabilitation effect for current robot systems and have great application potentials.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Efficient meta neural heuristic for multi-objective combinatorial optimization 23

Efficient meta neural heuristic for multi-objective combinat...

引用

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Jinbiao Chen Jiahai Wang Zizhen Zhang Zhiguang Cao Te Ye Siyuan Chen School of Computer Science and Engineering Sun Yat-sen University P.R. China School of Computer Science and Engineering Sun Yat-sen University P.R. China and Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education Sun Yat-sen University P.R. China and Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou P.R. China School of Computing and Information Systems Singapore Management University Singapore

Recently, neural heuristics based on deep reinforcement learning have exhibited promise in solving multi-objective combinatorial optimization problems (MOCOPs). However, they are still struggling to achieve high learning efficiency and solution quality. To tackle this issue, we propose an efficient meta neural heuristic (EMNH), in which a meta-model is first trained and then fine-tuned with a few steps to solve corresponding single-objective subproblems. Specifically, for the training process, a (partial) architecture-shared multi-task model is leveraged to achieve parallel learning for the meta-model, so as to speed up the training; meanwhile, a scaled symmetric sampling method with respect to the weight vectors is designed to stabilize the training. For the fine-tuning process, an efficient hierarchical method is proposed to systematically tackle all the subproblems. Experimental results on the multi-objective traveling salesman problem (MOTSP), multi-objective capacitated vehicle routing problem (MOCVRP), and multi-objective knapsack problem (MOKP) show that, EMNH is able to outperform the state-of-the-art neural heuristics in terms of solution quality and learning efficiency, and yield competitive solutions to the strong traditional heuristics while consuming much shorter time.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Super-Resolution-Based Channel Completion for Massive MISO Systems

引用

IEEE Signal Processing Letters 2025年 32卷 2254-2258页

作者： Zu, Keke He, Yuhan Chen, Hongyang Zheng, Yu Haardt, Martin Yangtze Delta Region Institute (Quzhou) University of Electronic Science and Technology of China Zhejiang China Shenzhen Key Laboratory of Advanced Machine Learning and Applications Shenzhen University Shenzhen China Research Center for Graph Computing Zhejiang Lab Hangzhou China JD Intelligent Cities Research Beijing China Communications Research Lab Ilmenau University of Technology Ilmenau Germany

With the deployment of large-scale antenna arrays, the already limited time-frequency resources are becoming increasingly scarce. In this study, we propose a novel Laplacian Pyramid Channel Completion Network (LPCCNet) designed for channel completion, thereby reducing the demand for time-frequency resources in massive MIMO systems. Compared with existing network models, the proposed LPCCNet, by employing a progressive upsampling architecture, effectively mitigates aliasing effects, suppresses error propagation, and achieves a substantial reduction in computational complexity. The simulation results show that LPCCNet achieves a superior channel completion quality compared to existing methods, particularly in rapidly time-varying scenarios. © 1994-2012 IEEE.

关键词： Channel estimation Feature extraction Image reconstruction Interpolation Training Superresolution Filters Vectors Convolutional neural networks Time-frequency analysis

来源：评论

学校读者我要写书评

暂无评论

Continual learning of new diseases with dual distillation and ensemble strategy 23rd

Continual learning of new diseases with dual distillation an...

引用

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

作者： Li, Zhuoyun Zhong, Changhong Wang, Ruixuan Zheng, Wei-Shi School of Data and Computer Science Sun Yat-sen University Guangzhou China Key Laboratory of Machine Intelligence and Advanced Computing MOE Guangzhou China Pazhou Lab Guangzhou China

ISBN: (纸本)9783030597092

Most intelligent diagnosis systems are developed for one or a few specific diseases, while medical specialists can diagnose all diseases of certain organ or tissue. Since it is often difficult to collect data of all diseases, it would be desirable if an intelligent system can initially diagnose a few diseases, and then continually learn to diagnose more and more diseases with coming data of these new classes in the future. However, current intelligent systems are characterised by catastrophic forgetting of old knowledge when learning new classes. In this paper, we propose a new continual learning framework to alleviate this issue by simultaneously distilling both old knowledge and recently learned new knowledge and by ensembling the class-specific knowledge from the previous classifier and the learned new classifier. Experiments showed that the proposed method outperforms state-of-the-art methods on multiple medical and natural image datasets. © Springer Nature Switzerland AG 2020.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

A Clinical-oriented Multi-level Contrastive Learning Method for Disease Diagnosis in Low-quality Medical Images

arXiv

引用

arXiv 2024年

作者： Hou, Qingshan Cheng, Shuai Cao, Peng Yang, Jinzhu Liu, Xiaoli Zaiane, Osmar R. Tham, Yih Chung Computer Science and Engineering Northeastern University Shenyang China Key Laboratory of Intelligent Computing in Medical Image Ministry of Education Northeastern University Shenyang China National Frontiers Science Center for Industrial Intelligence and Systems Optimization Shenyang110819 China Alberta Machine Intelligence Institute University of Alberta Edmonton Canada Yong Loo Lin School of Medicine National University of Singapore Singapore

Representation learning offers a conduit to elucidate distinctive features within the latent space and interpret the deep models. However, the randomness of lesion distribution and the complexity of low-quality factors in medical images pose great challenges for models to extract key lesion features. Disease diagnosis methods guided by contrastive learning (CL) have shown significant advantages in lesion feature representation. Nevertheless, the effectiveness of CL is highly dependent on the quality of the positive and negative sample pairs. In this work, we propose a clinical-oriented multi-level CL framework that aims to enhance the model's capacity to extract lesion features and discriminate between lesion and low-quality factors, thereby enabling more accurate disease diagnosis from low-quality medical images. Specifically, we first construct multi-level positive and negative pairs to enhance the model's comprehensive recognition capability of lesion features by integrating information from different levels and qualities of medical images. Moreover, to improve the quality of the learned lesion embeddings, we introduce a dynamic hard sample mining method based on self-paced learning. The proposed CL framework is validated on two public medical image datasets, EyeQ and Chest X-ray, demonstrating superior performance compared to other state-of-the-art disease diagnostic methods. Copyright © 2024, The Authors. All rights reserved.

关键词： Image enhancement

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：