检索结果-内蒙古大学图书馆

Towards Feature Distribution Alignment and Diversity Enhancement for data-Free Quantization

学校读者我要写书评

暂无评论

Towards Feature Distribution Alignment and Diversity Enhance...

IEEE International Conference on data Mining (ICDM)

作者： Yangcheng Gao Zhao Zhang Richang Hong Haijun Zhang Jicong Fan Shuicheng Yan School of Computer Science and Information Engineering Hefei University of Technology Hefei China Key Laboratory of Knowledge Engineering with Big Data (Ministry of Education) & Intelligent Interconnected Systems Laboratory of Anhui Province Hefei University of Technology Hefei China Department of Computer Science Harbin Institute of Technology (Shenzhen) Xili University Town Shenzhen China School of Data Science The Chinese University of Hong Kong Shenzhen China Shenzhen Research Institute of Big Data Shenzhen China National University of Singapore Singapore

To obtain lower inference latency and less memory footprint of deep neural networks, model quantization has been widely employed in deep model deployment, by converting the floating points to low-precision integers. However, previous methods (such as quantization aware training and post training quantization) require original data for the fine-tuning or calibration of quantized model, which makes them inapplicable to the cases that original data are not accessed due to privacy or security. This gives birth to the data-free quantization method with synthetic data generation. While current data-free quantization methods still suffer from severe performance degradation when quantizing a model into lower bit, caused by the low inter-class separability of semantic features. To this end, we propose a new and effective data-free quantization method termed ClusterQ, which utilizes the feature distribution alignment for synthetic data generation. To obtain high inter-class separability of semantic features, we cluster and align the feature distribution statistics to imitate the distribution of real data, so that the performance degradation is alleviated. Moreover, we incorporate the diversity enhancement to solve class-wise mode collapse. We also employ the exponential moving average to update the centroid of each cluster for further feature distribution improvement. Extensive experiments based on different deep models (e.g., ResNet-18 and MobileNet-V2) over the ImageNet dataset demonstrate that our proposed ClusterQ model obtains state-of-the-art performance.

关键词： Degradation Training Quantization (signal) Semantics Neural networks Memory management data models

FineFormer: Fine-Grained Adaptive Object Transformer for Image Captioning

学校读者我要写书评

暂无评论

FineFormer: Fine-Grained Adaptive Object Transformer for Ima...

IEEE International Conference on data Mining (ICDM)

作者： Bo Wang Zhao Zhang Jicong Fan Mingbo Zhao Choujun Zhan Mingliang Xu School of Computer Science and Information Engineering Hefei University of Technology Hefei China Key Laboratory of Knowledge Engineering with Big Data (Ministry of Education) & Intelligent Interconnected Systems Laboratory of Anhui Province Hefei University of Technology Hefei China School of Data Science The Chinese University of Hong Kong Shenzhen China Shenzhen Research Institute of Big Data Shenzhen China School of Information Science and Technology Donghua University Shanghai China School of Electrical and Computer Engineering Nanfang College Guangzhou Guangzhou China School of Information Engineering Zhengzhou University Zhengzhou China

Image captioning is still a challenging task aiming at describing the contents of image by words. Current image caption methods usually assume the object relation to be important if the semantic and spatial geometric relationships between objects are close and large, but the relations meeting this assumption are not necessarily important to describe the contents of image in a fine-grained way. That is, the importance of fine-grained object relations is not properly taken into account. Besides, current Transformer based image caption models also fail to consider the importance of fine-grained objects, since they generate all the words of a sentence at one time, which cannot Figure out which objects are more important and vice versa. In this paper, we propose a novel Fine-grained Adaptive Object Transformer (FineFormer) network, which can jointly discover the importance of fine-grained objects and object relations for image captioning. Specifically, a new concept of adaptive soft-foreground attention is proposed to highlight the fine-grained objects dominating the descriptive contents. To characterize and calculate the important relations between fine-grained objects, we also propose an adaptive object relation attention to refine the object relation from the generation process of relation. As such, FineFormer can describe the contents of image more accurately, by reducing the interference of unimportant objects in the background. Extensive experiments on the highly-competitive MS-COCO dataset demonstrated the superiority of our FineFormer.

关键词： Adaptation models Adaptive systems Current transformers Semantics Interference data mining Task analysis

Sentiment Analysis of Semester Learning Essays in Design Education

学校读者我要写书评

暂无评论

Sentiment Analysis of Semester Learning Essays in Design Edu...

IEEE International Conference on Industrial engineering and engineering Management

作者： Z. H. Wang Z. J. Ming G. X. Wang F. Mistree J. K. Allen School of Mechanical Engineering Beijing Institute of Technology Beijing China Yangtze Delta Region Academy of Beijing Institute of Technology Jiaxing China Key Laboratory of Industry Knowledge & Data Fusion Technology and Application Ministry of Industry and Information Technology Beijing China The Systems Realization Laboratory The University of Oklahoma Norman United States

AME4163: Principles of engineering Design is a design, build and test course offered at the University of Oklahoma, Norman, USA. Throughout the semester students are expected to reflect on authentic and immersive experiences and write Take-aways and Learning Statements. At the end of the semester each student submits a summative Semester Learning Essay. Between 2019–2021 we collected about 10,000 Take-aways and Learning Statements from the Semester Learning Essays of nearly 400 students. In this paper, we attempt to help instructors gain insight of what students have learned through these materials and thence improve the delivery of the course in the future. We analyze their Take-aways from summative Semester Learning Essays by using dictionary-based sentiment analysis to assess students' subjective feelings toward what they have learned. The results prove that we gain insight into their learning thereby providing instructors with evidence-based guidance on modifying the course from different perspectives. The method proposed is generalizable to courses that involve authentic immersive experiences.

关键词：

CLDG: Contrastive Learning on Dynamic Graphs

学校读者我要写书评

暂无评论

CLDG: Contrastive Learning on Dynamic Graphs

International Conference on data engineering

作者： Yiming Xu Bin Shi Teng Ma Bo Dong Haoyi Zhou Qinghua Zheng Department of Computer Science and Technology Xi’an Jiaotong University China Shaanxi Provincial Key Laboratory of Big Data Knowledge Engineering Xi’an Jiaotong University China Department of Distance Education Xi’an Jiaotong University China School of Software Beihang University China Advanced Innovation Center for Big Data and Brain Computing Beihang University China

The graph with complex annotations is the most potent data type, whose constantly evolving motivates further exploration of the unsupervised dynamic graph representation. One of the representative paradigms is graph contrastive learning. It constructs self-supervised signals by maximizing the mutual information between the statistic graph’s augmentation views. However, the semantics and labels may change within the augmentation process, causing a significant performance drop in downstream tasks. This drawback becomes greatly magnified on dynamic graphs. To address this problem, we designed a simple yet effective framework named CLDG. Firstly, we elaborate that dynamic graphs have temporal translation invariance at different levels. Then, we proposed a sampling layer to extract the temporally-persistent signals. It will encourage the node to maintain consistent local and global representations, i.e., temporal translation invariance under the timespan views. The extensive experiments demonstrate the effectiveness and efficiency of the method on seven datasets by outperforming eight unsupervised state-of-the-art baselines and showing competitiveness against four semi-supervised methods. Compared with the existing dynamic graph method, the number of model parameters and training time is reduced by an average of 2,001.86 times and 130.31 times on seven datasets, respectively. The code and data are available at: https://***/yimingxu24/CLDG.

关键词：

Online Robust Lagrangian Support Vector Machine against Adversarial Attack

学校读者我要写书评

暂无评论

Procedia Computer Science 2018年 139卷 173-181页

作者： Yue Ma Yiwei He Yingjie Tian School of Mathematical Sciences University of Chinese Academy of Sciences Beijing 100049 China Research Center on Fictitious Economy and Data Science Chinese Academy of Sciences Beijing 100190 China School of Computer and Control Engineering University of Chinese Academy of Sciences Beijing 100049 China Key Laboratory of Big Data Mining and Knowledge management Beijing 100190 China School of Economics and Management University of Chinese Academy of Sciences Beijing 100190 China

In adversarial environment such as intrusion detection and spam filtering, the adversary-intruder or spam advertiser may attempt to produce contaminate training instance and manipulate the learning of classifier. In order to keep good classification performance, many robuster learning methods have been proposed to deal with the adversarial attack. Support Vector Machines(SVMs) is a kind of successful approach in the adversarial classification tasks and the investigation of robust SVMs is very popular. However, in many real application, the data including stain instance is coming dynamically. Batch learning which needs retraining when encountering new samples, will consume more computing resources. In this paper, we propose a robust Lagrangian support vector machine (RLSVM) with modified kernel matrix and explore the online learning algorithm on it. The experimental results show the robustness of RLSVM against label noise produced by adversaries under the online adversarial environment.

关键词： adversarial attack poison attack label noise online learning Lagrangian SVM

SVT-Net: Super Light-Weight Sparse Voxel Transformer for Large Scale Place Recognition

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Fan, Zhaoxin Song, Zhenbo Liu, Hongyan Lu, Zhiwu He, Jun Du, Xiaoyong Key Laboratory of Data Engineering Knowledge Engineering of MOE School of Information Renmin University of China Beijing100872 China Gaoling School of Artificial Intelligence Renmin University of China Beijing100872 China School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing210094 China Department of Management Science and Engineering Tsinghua University Beijing100084 China

Simultaneous Localization and Mapping (SLAM) and Autonomous Driving are becoming increasingly more important in recent years. Point cloud-based large scale place recognition is the spine of them. While many models have been proposed and have achieved acceptable performance by learning short-range local features, they always skip long-range contextual properties. Moreover, the model size also becomes a serious shackle for their wide applications. To overcome these challenges, we propose a super light-weight network model termed SVT-Net. On top of the highly efficient 3D Sparse Convolution (SP-Conv), an Atom-based Sparse Voxel Transformer (ASVT) and a Cluster-based Sparse Voxel Transformer (CSVT) are proposed respectively to learn both short-range local features and long-range contextual features. Consisting of ASVT and CSVT, SVT-Net can achieve state-of-the-art performance in terms of both recognition accuracy and running speed with a super-light model size (0.9M parameters). Meanwhile, for the purpose of further boosting efficiency, we introduce two simplified versions, which also achieve state-of-the-art performance and further reduce the model size to 0.8M and 0.4M respectively. Copyright © 2021, The Authors. All rights reserved.

关键词： Machine learning

CLDG: Contrastive Learning on Dynamic Graphs

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xu, Yiming Shi, Bin Ma, Teng Dong, Bo Zhou, Haoyi Zheng, Qinghua Department of Computer Science and Technology Xi’an Jiaotong University China Shaanxi Provincial Key Laboratory of Big Data Knowledge Engineering Xi’an Jiaotong University China Department of Distance Education Xi’an Jiaotong University China School of Software Beihang University China Advanced Innovation Center for Big Data and Brain Computing Beihang University China

关键词： Contrastive Learning

Asymmetric Cross-Scale Alignment for Text-Based Person Search

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Ji, Zhong Hu, Junhua Liu, Deyin Wu, Lin Yuanbo Zhao, Ye The School of Electrical and Information Engineering Tianjin University Tianjin300072 China Anhui Provincial Key Laboratory of Multimodal Cognitive Computation School of Artificial Intelligence Anhui University Hefei230039 China Department of Computer Science Swansea University SA1 8EN United Kingdom The Key Laboratory of Knowledge Engineering with Big Data School of Computer Science and Information Engineering Hefei University of Technology Hefei230009 China

Text-based person search (TBPS) is of significant importance in intelligent surveillance, which aims to retrieve pedestrian images with high semantic relevance to a given text description. This retrieval task is characterized with both modal heterogeneity and fine-grained matching. To implement this task, one needs to extract multi-scale features from both image and text domains, and then perform the cross-modal alignment. However, most existing approaches only consider the alignment confined at their individual scales, e.g., an image-sentence or a region-phrase scale. Such a strategy adopts the presumable alignment in feature extraction, while overlooking the cross-scale alignment, e.g., image-phrase. In this paper, we present a transformer-based model to extract multi-scale representations, and perform Asymmetric Cross-Scale Alignment (ACSA) to precisely align the two modalities. Specifically, ACSA consists of a global-level alignment module and an asymmetric cross-attention module, where the former aligns an image and texts on a global scale, and the latter applies the cross-attention mechanism to dynamically align the cross-modal entities in region/image-phrase scales. Extensive experiments on two benchmark datasets CUHK-PEDES and RSTPReid demonstrate the effectiveness of our approach. Codes are available at https://***/mulhjh/ACSA. Copyright © 2022, The Authors. All rights reserved.

关键词： Alignment

Fast One-Stage Unsupervised Domain Adaptive Person Search

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Cui, Tianxiang Wang, Huibing Peng, Jinjia Deng, Ruoxi Fu, Xianping Wang, Yang The College of Information and Science Technology Dalian Maritime University China The College School of Cyber Security and Computer Hebei University China The College of Computer Science and Artificial Intelligence Wenzhou University China Key Laboratory of Knowledge Engineering with Big Data Hefei University of Technology China

Unsupervised person search aims to localize a particular target person from a gallery set of scene images without annotations, which is extremely challenging due to the unexpected variations of the unlabeled domains. However, most existing methods dedicate to developing multi-stage models to adapt domain variations while using clustering for iterative model training, which inevitably increases model complexity. To address this issue, we propose a Fast One-stage Unsupervised person Search (FOUS) which complementary integrates domain adaptaion with label adaptaion within an end-to-end manner without iterative clustering. To minimize the domain discrepancy, FOUS introduced an Attention-based Domain Alignment Module (ADAM) which can not only align various domains for both detection and ReID tasks but also construct an attention mechanism to reduce the adverse impacts of low-quality candidates resulting from unsupervised detection. Moreover, to avoid the redundant iterative clustering mode, FOUS adopts a prototype-guided labeling method which minimizes redundant correlation computations for partial samples and assigns noisy coarse label groups efficiently. The coarse label groups will be continuously refined via label-flexible training network with an adaptive selection strategy. With the adapted domains and labels, FOUS can achieve the state-of-the-art (SOTA) performance on two benchmark datasets, CUHK-SYSU and PRW. The code is available at https://***/whbdmu/FOUS © 2024, CC0.

关键词： Iterative methods