检索结果-内蒙古大学图书馆

Human-Centric Transformer for Domain Adaptive Action Recognition

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lin, Kun-Yu Zhou, Jiaming Zheng, Wei-Shi The School of Computer Science and Engineering Sun Yat-sen University Guangzhou510275 China The Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education Guangzhou510275 China The Guangdong Province Key Laboratory of Information Security Technology Sun Yat-sen University Guangzhou510275 China The Peng Cheng Laboratory Shenzhen518005 China

We study the domain adaptation task for action recognition, namely domain adaptive action recognition, which aims to effectively transfer action recognition power from a label-sufficient source domain to a label-free target domain. Since actions are performed by humans, it is crucial to exploit human cues in videos when recognizing actions across domains. However, existing methods are prone to losing human cues but prefer to exploit the correlation between non-human contexts and associated actions for recognition, and the contexts of interest agnostic to actions would reduce recognition performance in the target domain. To overcome this problem, we focus on uncovering human-centric action cues for domain adaptive action recognition, and our conception is to investigate two aspects of human-centric action cues, namely human cues and human-context interaction cues. Accordingly, our proposed Human-Centric Transformer (HCTransformer) develops a decoupled human-centric learning paradigm to explicitly concentrate on human-centric action cues in domain-variant video feature learning. Our HCTransformer first conducts human-aware temporal modeling by a human encoder, aiming to avoid a loss of human cues during domain-invariant video feature learning. Then, by a Transformer-like architecture, HCTransformer exploits domain-invariant and action-correlated contexts by a context encoder, and further models domain-invariant interaction between humans and action-correlated contexts. We conduct extensive experiments on three benchmarks, namely UCF-HMDB, Kinetics-NecDrone and EPIC-Kitchens-UDA, and the state-of-the-art performance demonstrates the effectiveness of our proposed HCTransformer. Copyright © 2024, The Authors. All rights reserved.

关键词： Benchmarking

LinNet: linear network for efficient point cloud representation learning 24

学校读者我要写书评

暂无评论

LinNet: linear network for efficient point cloud representat...

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Hao Deng Kunlei Jing Shengmei Cheng Cheng Liu Jiawei Ru Jiang Bo Lin Wang State-Province Joint Engineering and Research Center of Advanced Networking and Intelligent Information Services School of Information Science and Technology Northwest University and Shaanxi Key Laboratory of Higher Education Institution of Generative Artificial Intelligence and Mixed Reality School of Software Engineering Xi'an Jiaotong University and Department of Computing The Hong Kong Polytechnic University

ISBN: (纸本)9798331314385

Point-based methods have made significant progress, but improving their scalability in large-scale 3D scenes is still a challenging problem. In this paper, we delve into the point-based method and develop a simpler, faster, stronger variant model, dubbed as LinNet. In particular, we first propose the disassembled set abstraction (DSA) module, which is more effective than the previous version of set abstraction. It achieves more efficient local aggregation by leveraging spatial anisotropy and channel anisotropy separately. Additionally, by mapping 3D point clouds onto 1D space-filling curves, we enable parallelization of down-sampling and neighborhood queries on GPUs with linear complexity. LinNet, as a purely point-based method, outperforms most previous methods in both indoor and outdoor scenes without any extra attention, and sparse convolution but merely relying on a simple MLP. It achieves the mIoU of 73.7%, 81.4%, and 69.1% on the S3DIS Area5, NuScenes, and SemanticKITTI validation benchmarks, respectively, while speeding up almost 10x times over PointNeXt. Our work further reveals both the efficacy and efficiency potential of the vanilla point-based models in large-scale representation learning. Our code will be available at https://***/DengH293/LinNet.

关键词：

WINDFORMER: BI-DIRECTIONAL LONG-DISTANCE SPATIO-TEMPORAL NETWORK FOR WIND SPEED PREDICTION

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Xuewei Shang, Zewen Liu, Zhiqiang Yu, Jian Xiong, Wei Yu, Mei College of Intelligence and Computing Tianjin University Tianjin300350 China Tianjin Key Laboratory of Cognitive Computing and Application Tianjin300350 China Tianjin Key Laboratory of Advanced Networking Tianjin300350 China Tianjin International Engineering Institute Tianjin University Tianjin300350 China TCU School of Civil Engineering Tianjin Chengjian University Tianjin300384 China

Wind speed prediction is critical to the management of wind power generation. Due to the large range of wind speed fluctuations and wake effect, there may also be strong correlations between long-distance wind turbines. This difficult-to-extract feature has become a bottleneck for improving accuracy. History and future time information includes the trend of airflow changes, whether this dynamic information can be utilized will also affect the prediction effect. In response to the above problems, this paper proposes Windformer. First, Windformer divides the wind turbine cluster into multiple non-overlapping windows and calculates correlations inside the windows, then shifts the windows partially to provide connectivity between windows, and finally fuses multi-channel features based on detailed and global information. To dynamically model the change process of wind speed, this paper extracts time series in both history and future directions simultaneously. Compared with other current-advanced methods, the Mean Square Error (MSE) of Windformer is reduced by 0.5% to 15% on two datasets from NERL. Copyright © 2023, The Authors. All rights reserved.

关键词： Mean square error

Homophily Induced Contrastive Attributed Graph Clustering

学校读者我要写书评

暂无评论

IEEE Transactions on Circuits and Systems for Video Technology 2025年

作者： Chen, Man-Sheng Lai, Pei-Yuan Liao, De-Zhang Wang, Chang-Dong Lai, Jian-Huang Sun Yat-sen University School of Computer Science and Engineering Guangzhou China Guangxi Zhuang Autonomous Region Information Center Guangxi Key Laboratory of Digital Infrastructure Nanning China Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China South China Technology Commercialization Center Guangzhou China Guangdong Key Laboratory of Information Security Technology Guangzhou China

Attributed graph clustering, aiming to discover the underlying graph structure and partition the graph nodes into several disjoint categories, is a basic task in graph data analysis. Although recent efforts over graph contrastive clustering have achieved decent performances, most of them get accustomed to construct the positive neighbor set by the generated pseudo clustering information, directly ignoring the ready-made neighbor nodes and the underlying semantics of edges in a graph. How to well deal with the graph neighbor-specific information to facilitate the performance of graph contrastive clustering is still a challenging problem. Therefore, in this work, we propose a novel Homophily Induced Contrastive Attributed Graph Clustering (HomoCAGC) method, where the power of homophily is exploited in facilitating the performance of contrastive attributed graph clustering, while the pseudo homophily in a graph is also explored and distinguished. Especially, the node feature as reliable information guidance is used to compute the underlying feature-oriented pair-wise node similarity, based on which the positive node pairs in contrastive regularizer are adjusted for better node representation characterization. According to the refined node representations, a triplet self-supervised clustering objective is well-designed to ensure the output embedding is cluster-oriented, and suitable for the downstream clustering task. Extensive experiments on seven benchmarks are conducted to demonstrate the effectiveness of HomoCAGC. © 1991-2012 IEEE.

关键词： Attributed graph clustering contrastive learning graph homophily self-supervised clustering

Statistical physics of deep neural networks: Initialization toward optimal channels

学校读者我要写书评

暂无评论

Physical Review Research 2023年第2期5卷 023023-023023页

作者： Kangyu Weng Aohua Cheng Ziyang Zhang Pei Sun Yang Tian Tsien Excellence in Engineering Program Tsinghua University Beijing 100084 China Laboratory of Advanced Computing and Storage Central Research Institute 2012 Laboratories Huawei Technologies Co. Ltd. Beijing 100084 China Department of Psychology & Tsinghua Laboratory of Brain and Intelligence Tsinghua University Beijing 100084 China

In deep learning, neural networks serve as noisy channels between input data and its latent representation. This perspective naturally relates deep learning with the pursuit of constructing channels with optimal performance in information transmission and representation. While considerable efforts are concentrated on realizing optimal channel properties during network optimization, we study a frequently overlooked possibility that neural networks can be initialized toward optimal channels. Our theory, consistent with experimental validation, identifies primary mechanics underlying this unknown possibility and suggests intrinsic connections between statistical physics and deep learning. Unlike the conventional theories that characterize neural networks applying the classic mean-field approximation, we offer analytic proof that this extensively applied simplification scheme is not appropriate in studying neural networks as information channels. To fill this gap, we develop a restricted mean-field framework applicable for characterizing the limiting behaviors of information propagation in neural networks without strong assumptions on inputs. Based on it, we propose an analytic theory to prove that mutual information maximization is realized between inputs and propagated signals when neural networks are initialized at dynamic isometry, a case where information transmits via norm-preserving mappings. These theoretical predictions are validated by experiments on real neural networks, suggesting the robustness of our theory against finite-size effects. Finally, we analyze our findings with information bottleneck theory to confirm the precise relations among dynamic isometry, mutual information maximization, and optimal channel properties in deep learning. Our work may lay a cornerstone for promoting deep learning in terms of network initialization and suggest general statistical physics mechanisms underlying diverse deep learning techniques.

关键词： Neuronal dynamics Artificial neural networks Deep learning Information theory Random matrix theory

Accurately Identifying Coronary Atherosclerotic Heart Disease through Merged Beats of Electrocardiogram

学校读者我要写书评

暂无评论

Accurately Identifying Coronary Atherosclerotic Heart Diseas...

2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022

作者： Wang, Xinfeng Qi, Mengling Dong, Chengzhi Zhang, Haoyang Yang, Yuedong Zhao, Huiying Sun Yat-sen University Key Laboratory of Machine Intelligence and Advanced Computing of Moe School of Computer Science and Engineering Guangzhou China Sun Yat-sen Memorial Hospital Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and Gene Regulation Department of Medical Research Center Guangzhou China Sun Yat-sen University School of Life Sciences Guangzhou China

ISBN: (纸本)9781665468190

Coronary Atherosclerotic Heart Disease (CAHD) is one kind of severe heart disease that is the dominating cause of death from non-communicable diseases worldwide. CAHD can be early detected through pre-symptomatic health check-ups, and the electrocardiogram (ECG) is common for non-invasive health check diagnoses. Traditionally, ECG signals are utilized to extract clinical features that are then input into machine learning methods for training and prediction. While these extracted features are interpretable, they are difficult to break through known features. On the other hand, ECG can be directly input to deep learning techniques, but such methods are usually limited by small sample sizes. Here, we propose to merge multiple beats of raw signal into one beat, which greatly reduces the complexity while maintaining the raw information. Moreover, we have constructed the largest benchmark dataset for 1113 CAHD patients of 12-lead ECG signals from the UK Biobank database and used the data to train a deep learning model. The results indicated that merged beat signals could achieve the best performance corresponding to an AUC of 0.71 and accuracy of 0.7, which is 4% higher than models using the raw signals and 6% higher than those using the clinical features. Further intuitive interpretation revealed that ST waves in lead II and V3 are the most closely associated with CAHD, consistent with clinical observations. © 2022 IEEE.

关键词： Electrocardiograms

cRVR: A Cache-Friendly Approach to Enhancing Request Privacy for Online Video Services

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhang, Xianzhi Xiao, Linchang Zhou, Yipeng Hu, Miao Wu, Di Lui, John C.S. Sheng, Quan Z. School of Computer Science and Engineering Sun Yat-sen University Guangzhou510006 China The Key Laboratory of Machine Intelligence and Advanced Computing Sun Yat-sen University Ministry of Education China School of Computing Faculty of Science and Engineering Macquarie University 2122 Australia Department of Computer Science & Engineering Chinese University of Hong Kong Hong Kong

—As users conveniently stream their favorite online videos, video request records are automatically stored by video content providers, which have a high chance of privacy leakage. Unfortunately, most existing privacy-enhancing approaches are not applicable for protecting user privacy in video requests, because they cannot be easily altered or distorted by users and must be visible for content providers to stream correct videos. To preserve request privacy in online video services, it is possible to request additional videos that are irrelevant to users’ interests so that content providers cannot precisely infer users’ interest information. However, a naive redundant requesting approach would significantly degrade the performance of edge caches and increase bandwidth overhead. In this paper, we are among the first to propose a Cache-Friendly Redundant Video Requesting (cRVR) algorithm for User Devices (UDs) and its corresponding caching algorithm for the Edge Cache (EC), which can effectively mitigate the problem of request privacy leakage with minimal impact on the EC’s performance. To tackle the problem, we first develop a Stackelberg game to analyze the dedicated interaction between UDs and EC, and obtain their optimal strategies to maximize their respective utility. For UDs, the utility function is a combination of both video playback utility and privacy protection utility. We prove the existence and uniqueness of the equilibrium of the Stackelberg game. Extensive experiments are conducted with real traces to demonstrate that cRVR can effectively protect video request privacy by reducing up to 59.03% of privacy disclosure compared to baseline algorithms. Meanwhile, the caching performance of EC is only slightly affected. Copyright © 2023, The Authors. All rights reserved.

关键词： Differential privacy

A Neuroinspired Contrast Mechanism Enables Few-Shot Object Detection

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Yang, Lingxiao Chen, Dapeng Chen, Yifei Peng, Wei Xie, Xiaohua School of Computer Science and Engineering Sun Yat-sen University Guangzhou China Guangdong Province Key Laboratory of Information Security Technology Guangzhou China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education Guangzhou China Huawei Technlogies Shenzhen China

Object detectors based on deep neural networks often require a large number of annotated images for training. In the most realistic scenario, only a few training data are available;thus, Few-Shot Object Detection (FSOD) has become an attractive research topic. In this problem setting, two-stage proposal-based Faster R-CNN detectors are widely used in multiple previous studies, while its performance still lags far behind the one trained with lots of data. We identify the main bottleneck of training a detector in the few-shot scenario as the insufficient number of samples for discovering robust comparison-based relationships (e.g., proposal-level and category-level relationships). Moreover, we found that existing methods explore these relationships only during the training phase. It is interesting to investigate whether we can effectively incorporate such relationships during model training and testing to improve FSOD. To this end, we borrow the idea of "contrast" from neuroscience studies, and propose a simple yet effective architecture, named Residual Contrast Faster R-CNN (RcFRCN). Our RcFRCN extends the existing state-of-the-art FSOD detector by introducing two newly designed contrast blocks, i.e., a Residual Spatial Contrast Block and a Residual Proposal Contrast Block. The former block mainly explores cross-spatial contrast information in a feature map extracted by the backbone network, and the latter block discovers contrast cues across proposal features outputted by the RPN module. As a result, these blocks store differences and commodities in different perspectives and are used in both training and test phases, thus enhancing the ability of transfer learning. Extensive experiments demonstrate the effectiveness of our RcFRCN. Moreover, our method also helps the detector to retain more previous knowledge and obtains better performance in the detection of base classes. © 2023, The Authors. All rights reserved.

关键词： Object detection

Adapter Learning in Pretrained Feature Extractor for Continual Learning of Diseases

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhang, Wentao Huang, Yujun Zhang, Tong Zou, Qingsong Zheng, Wei-Shi Wang, Ruixuan School of Computer Science and Engineering Sun Yat-sen Univerisity Guangzhou China Peng Cheng Laboratory Shenzhen China Guangdong Province Key Laboratory of Computational Science Sun Yat-sen University Guangzhou China Key Laboratory of Machine Intelligence and Advanced Computing MOE Guangzhou China

Currently intelligent diagnosis systems lack the ability of continually learning to diagnose new diseases once deployed, under the condition of preserving old disease knowledge. In particular, updating an intelligent diagnosis system with training data of new diseases would cause catastrophic forgetting of old disease knowledge. To address the catastrophic forgetting issue, an Adapter-based Continual Learning framework called ACL is proposed to help effectively learn a set of new diseases at each round (or task) of continual learning, without changing the shared feature extractor. The learnable lightweight task-specific adapter(s) can be flexibly designed (e.g., two convolutional layers) and then added to the pretrained and fixed feature extractor. Together with a specially designed task-specific head which absorbs all previously learned old diseases as a single ‘out-of-distribution’ category, task-specific adapter(s) can help the pretrained feature extractor more effectively extract discriminative features between diseases. In addition, a simple yet effective fine-tuning is applied to collaboratively fine-tune multiple task-specific heads such that outputs from different heads are comparable and consequently the appropriate classifier head can be more accurately selected during model inference. Extensive empirical evaluations on three image datasets demonstrate the superior performance of ACL in continual learning of new diseases. The source code is available at https: //***/GiantJun/CL_Pytorch. Copyright © 2023, The Authors. All rights reserved.

关键词： Diagnosis