检索结果-内蒙古大学图书馆

Learning Causal Representation for Training Cross-Domain Pose Estimator via Generative Interventions

学校读者我要写书评

暂无评论

Learning Causal Representation for Training Cross-Domain Pos...

International Conference on computer Vision (ICCV)

作者： Xiheng Zhang Yongkang Wong Xiaofei Wu Juwei Lu Mohan Kankanhalli Xiangdong Li Weidong Geng State Key Laboratory of CAD&CG College of Computer Science and Technology Zhejiang University School of Computing National University of Singapore Huawei Noah’s Ark Laboratory

ISBN: (纸本)9781665428132

3D pose estimation has attracted increasing attention with the availability of high-quality benchmark datasets. However, prior works show that deep learning models tend to learn spurious correlations, which fail to generalize beyond the specific dataset they are trained on. In this work, we take a step towards training robust models for cross-domain pose estimation task, which brings together ideas from causal representation learning and generative adversarial networks. Specifically, this paper introduces a novel framework for causal representation learning which explicitly exploits the causal structure of the task. We consider changing domain as interventions on images under the data-generation process and steer the generative model to produce counterfactual features. This help the model learn transferable and causal relations across different domains. Our framework is able to learn with various types of unlabeled datasets. We demonstrate the efficacy of our proposed method on both human and hand pose estimation task. The experiment results show the proposed approach achieves state-of-the-art performance on most datasets for both domain adaptation and domain generalization settings.

关键词： Training Representation learning Deep learning Visualization Three-dimensional displays Correlation Pose estimation

Towards high-accuracy and Low-latency Spiking Neural Networks with Two-Stage Optimization

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wang, Ziming Zhang, Yuhao Lian, Shuang Cui, Xiaoxin Yan, Rui Tang, Huajin College of Computer Science and Technology Zhejiang University Hangzhou310027 China Research Center for Intelligent Computing Hardware Zhejiang Lab Hangzhou311100 China School of Integrated Circuits Peking University Beijing100871 China College of Computer Science Zhejiang University of Technology Hangzhou310014 China State Key Laboratory of Brain-Machine Intelligence Zhejiang University Hangzhou310027 China

Spiking neural networks (SNNs) operating with asynchronous discrete events show higher energy efficiency with sparse computation. A popular approach for implementing deep SNNs is ANN-SNN conversion combining both efficient training of ANNs and efficient inference of SNNs. However, the accuracy loss is usually non-negligible, especially under few time steps, which restricts the applications of SNN on latencysensitive edge devices greatly. In this paper, we first identify that such performance degradation stems from the misrepresentation of the negative or overflow residual membrane potential in SNNs. Inspired by this, we decompose the conversion error into three parts: quantization error, clipping error, and residual membrane potential representation error. With such insights, we propose a two-stage conversion algorithm to minimize those errors respectively. Besides, We show each stage achieves significant performance gains in a complementary manner. By evaluating on challenging datasets including CIFAR-10, CIFAR- 100 and ImageNet, the proposed method demonstrates the stateof- the-art performance in terms of accuracy, latency and energy preservation. Furthermore, our method is evaluated using a more challenging object detection task, revealing notable gains in regression performance under ultra-low latency when compared to existing spike-based detection algorithms. Codes are available at https://***/Windere/snn-cvt-dual-phase. Copyright © 2022, The Authors. All rights reserved.

关键词： Object detection

Mining Microbe-disease Interactions From Literature via a Transfer Learning Model

学校读者我要写书评

暂无评论

Research Square

Research Square 2021年

作者： Wu, Chengkun Xiao, Xinyi Yang, Canqun Chen, JinXiang Yi, Jiacai Qiu, Yanlong State Key Laboratory of High-Performance Computing National University of Defense Technology Changsha410073 China College of Computer National University of Defense Technology Changsha410073 China Department of General Surgery Xiangya Hospital Central South University Changsha410008 China

Background: Interactions of microbes and diseases are of great importance for biomedical research. However, large-scale curated databases for microbe-disease interactions are missing, as the amount of related literature is enormous and the curation process is costly and time-consuming. In this paper, we aim to construct a large-scale database for microbe-disease interactions automatically. We attained this goal via applying text mining methods based on a deep learning model with a moderate curation cost. We also built a user-friendly web interface to allow researchers navigate and query desired information. Results: For curation, we manually constructed a golden-standard corpora (GSC) and a sliver-standard corpora (SSC) for microbe-disease interactions. Then we proposed a text mining framework for microbe-disease interaction extraction without having to build a model from scratch. Firstly, we applied named entity recognition (NER) tools to detect microbe and disease mentions from texts. Then we transferred a deep learning model BERE to recognize relations between entities, which was originally built for drug-target interactions or drug-drug interactions. The introduction of SSC for model fine-tuning greatly improves the performance of detection for microbe-disease interactions, with an average reduction in error of approximately 10%. The resulting MDIDB website offers data browsing, custom search for specific diseases or microbes as well as batch download. Conclusions: Evaluation results demonstrate that our method outperform the baseline model (rule-based PKDE4J) with an average F1-score of 73.81%. For further validation, we randomly sampled nearly 1,000 predicted interactions by our model, and manually checked the correctness of each interaction, which gives a 73% accuracy. The MDIDB webiste is freely avaliable throuth http://***/index/ © 2021, CC BY.

关键词： Bacteria

FishEye8K: A Benchmark and Dataset for Fisheye Camera Object Detection

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Gochoo, Munkhjargal Otgonbold, Munkh-Erdene Ganbold, Erkhembayar Hsieh, Jun-Wei Chang, Ming-Ching Chen, Ping-Yang Dorj, Byambaa Al Jassmi, Hamad Batnasan, Ganzorig Alnajjar, Fady Abduljabbar, Mohammed Lin, Fang-Pang Department of Computer Science and Software Engineering United Arab Emirates University United Arab Emirates Emirates Center for Mobility Research United Arab Emirates University United Arab Emirates College of AI and Green Energy National Yang Ming Chiao Tung University Taiwan University at Albany State University of New York NY United States Department of Computer Science National Yang Ming Chiao Tung University Taiwan Mongolian University of Science and Technology Mongolia National Center for High-Performance Computing Taiwan

With the advance of AI, road object detection has been a prominent topic in computer vision, mostly using perspective cameras. Fisheye lens provides omnidirectional wide coverage for using fewer cameras to monitor road intersections, however with view distortions. To our knowledge, there is no existing open dataset prepared for traffic surveillance on fisheye cameras. This paper introduces an open FishEye8K benchmark dataset for road object detection tasks, which comprises 157K bounding boxes across five classes (Pedestrian, Bike, Car, Bus, and Truck). In addition, we present benchmark results of state-of-The-Art (SoTA) models, including variations of YOLOv5, YOLOR, YOLO7, and YOLOv8. The dataset comprises 8,000 images recorded in 22 videos using 18 fisheye cameras for traffic monitoring in Hsinchu, Taiwan, at resolutions of 1080×1080 and 1280×1280. The data annotation and validation process were arduous and time-consuming, due to the ultra-wide panoramic and hemispherical fisheye camera images with large distortion and numerous road participants, particularly people riding scooters. To avoid bias, frames from a particular camera were assigned to either the training or test sets, maintaining a ratio of about 70:30 for both the number of images and bounding boxes in each class. Experimental results show that YOLOv8 and YOLOR outperform on input sizes 640×640 and 1280×1280, respectively. The dataset will be available on the GitHub link with PASCAL VOC, MS COCO, and YOLO annotation formats. The Fish-Eye8K benchmark will provide significant contributions to the fisheye video analytics and smart city applications. Copyright © 2023, The Authors. All rights reserved.

关键词： Object detection

Application-level Benchmarking of Quantum computers using Nonlocal Game Strategies

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Furches, Jim Chehade, Sarah Hamilton, Kathleen Wiebe, Nathan Marrero, Carlos Ortiz College of Computing Georgia Institute of Technology AtlantaGA30332 United States Physical Detection Systems and Deployment Division Pacific Northwest National Laboratory RichlandWA99354 United States Quantum Computational Science Group Oak Ridge National Laboratory Oak RidgeTN37830 United States Department of Computer Science University of Toronto ONM5S 1A1 Canada High Performance Computing Group Pacific Northwest National Laboratory RichlandWA99354 United States Canadian Institute for Advanced Research TorontoONM5G 1M1 Canada Department of Electrical & Computer Engineering North Carolina State University RaleighNC27607 United States

In a nonlocal game, two noncommunicating players cooperate to convince a referee that they possess a strategy that does not violate the rules of the game. Quantum strategies allow players to optimally win some games by performing joint measurements on a shared entangled state, but computing these strategies can be challenging. We present a variational quantum algorithm to compute quantum strategies for nonlocal games by encoding the rules of a nonlocal game into a Hamiltonian. We show how this algorithm can generate a short-depth optimal quantum strategy for a graph coloring game with a quantum advantage. This quantum strategy is then evaluated on fourteen different quantum hardware platforms to demonstrate its utility as a benchmark. Finally, we discuss potential sources of errors that can explain the observed decreased performance of the executed task and derive an expression for the number of samples required to accurately estimate the win rate in the presence of noise. Copyright © 2023, The Authors. All rights reserved.

关键词： Hamiltonians

Contrastive Learning Based Human Activity Recognition Using Body Sensors

学校读者我要写书评

暂无评论

Contrastive Learning Based Human Activity Recognition Using ...

International Conference on Big Data computing and Communications (BIGCOM)

作者： Chuanshi Xie Yu Zhou Dandan Yu Shilong Sun Xiao Zhang College of Computer Science and Software Engineering Shenzhen University Shenzhen China Information Center The First Affiliated Hospital of Dalian Medical University Dalian China School of Mechanical Engineering and Automation Harbin Institute of Technology Shenzhen China Department of Computer Science South-Central Minzu University State Ethnic Affairs Commission Key Laboratory of Cyber-Physical Fusion Intelligent Computing South-Central Minzu University Wuhan China

Pervasive sensing and wearable sensor techniques have been increasingly employed to monitor and recognize human activities through body sensors in areas of smart healthcare and manufacturing. However, conventional machine learning or deep learning based human activity recognition (HAR) requires a large amount of labeled data, which is cost-expensive in real-world scenarios. To tackle this issue, we propose a contrastive learning framework for HAR (CL-HAR), which utilizes the generated unlabeled data as the input and explores the supervised features of the unlabeled data under the principle of self-supervised learning. A simple yet effective backbone network as a feature extractor for subsequent activity recognition is proposed. By using a small portion of the labeled samples as the training set which is fed into our learned feature extractor, we build a classifier and use the rest of the data to verify the feasibility and effectiveness of our CL method. Extensive experiments on three benchamark datasets and one real-world dataset demonstrate that CL-HAR can achieve better classification accuracy than compared supervised and semi-supervised methods with less labelled samples, which is of practical use.

关键词：

AdaFuse: Adaptive Medical Image Fusion Based on Spatial-Frequential Cross Attention

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Gu, Xianming Wang, Lihui Deng, Zeyu Cao, Ying Huang, Xingyu Zhu, Yue-Min Engineering Research Center of Text Computing & Cognitive Intelligence Ministry of Education Key Laboratory of Intelligent Medical Image Analysis and Precise Diagnosis of Guizhou Province State Key Laboratory of Public Big Data College of Computer Science and Technology Guizhou University Guiyang550025 China University Lyon INSA Lyon CNRS Inserm IRP Metislab CREATIS UMR5220 U1206 Lyon69621 France

Multi-modal medical image fusion is essential for the precise clinical diagnosis and surgical navigation since it can merge the complementary information in multi-modalities into a single image. The quality of the fused image depends on the extracted single modality features as well as the fusion rules for multi-modal information. Existing deep learning-based fusion methods can fully exploit the semantic features of each modality, they cannot distinguish the effective low and high frequency information of each modality and fuse them adaptively. To address this issue, we propose AdaFuse, in which multimodal image information is fused adaptively through frequency-guided attention mechanism based on Fourier transform. Specifically, we propose the cross-attention fusion (CAF) block, which adaptively fuses features of two modalities in the spatial and frequency domains by exchanging key and query values, and then calculates the cross-attention scores between the spatial and frequency features to further guide the spatial-frequential information fusion. The CAF block enhances the high-frequency features of the different modalities so that the details in the fused images can be retained. Moreover, we design a novel loss function composed of structure loss and content loss to preserve both low and high frequency information. Extensive comparison experiments on several datasets demonstrate that the proposed method outperforms state-of-the-art methods in terms of both visual quality and quantitative metrics. The ablation experiments also validate the effectiveness of the proposed loss and fusion strategy. © 2023, CC BY.

关键词： Deep learning

BTDetect: An Insider Threats Detection Approach Based on Behavior Traceability for IaaS Environments

学校读者我要写书评

暂无评论

BTDetect: An Insider Threats Detection Approach Based on Beh...

IEEE International Conference on Big Data and Cloud computing (BdCloud)

作者： Li Lin Shuang Li Xuhui Lv Bo Li College of Computer Science Beijing University of Technology Beijing China Beijing Key Laboratory of Trusted Computing Beijing China State Key Laboratory of Software Development Environment Beihang University Beijing China

How to detect malicious insiders’ improper access to tenant data has become more crucial in IaaS cloud environment, especially with the cloud administrators gaining more control on customers’ virtual machines and data in reality. In this paper, we propose an insider threats detection approach based on behavior traceability called BTDetect. First, we analyze the service invocation interfaces of IaaS cloud environment, such as computing service, remote call, management implementation and virtualization management, and condense the complete process of cloud user behavior. Using tree-based modeling technique, a behavior-tree construction algorithm is proposed to construct the normal behavior tree that can describe various legal operations of cloud users. Second , we set up trace points of cloud service behavior on multi-layer cloud service APIs, then we collect information of each interface being invoked across multiple nodes. Third, we use underlying virtualization behavior keyword matching technology to match the collected behaviors with the user's normal behavior tree and then the malicious internal threat can be identified through tree-based integrity analysis. Finally, some experiments are conducted to evaluate the feasibility and veracity of the proposed method in Openstack platform. The results suggest that our method can not only identify internal threat but also have high recognition rate.

关键词： Cloud computing Law Computational modeling Virtual machining Security Personnel Virtualization

Drastic increase of key rate in continuous-variable quantum key distribution using multimode states and non-Gaussian operations

学校读者我要写书评

暂无评论

Quantum Engineering

Quantum Engineering 2020年第3期2卷

作者： Huang, Anqi Institute for Quantum Information & State Key Laboratory of High Performance Computing College of Computer National University of Defense Technology Changsha China