检索结果-内蒙古大学图书馆

Disentangled features with direct sum decomposition for zero shot learning

NEUROCOMPUTING 2021年 426卷 216-226页

作者： Li, Bonan Han, Congying Guo, Tiande Zhao, Tong Univ Chinese Acad Sci Sch Math Sci 19A Yuquan Rd Beijing 100049 Peoples R China Chinese Acad Sci Key Lab Big Data Min & Knowledge Management 80 Zhonguancun East Rd Beijing 100190 Peoples R China

The core problem of zero-shot learning (ZSL) is extracting concise feature from cumbersome and intricate semantic space. While semantic vectors were mainly used to constrain the distance between different classes in embedding space in existing ZSL methods, we first emphasize the importance of the relationship between different dimensions in semantic space. In this paper, we propose a novel inductive approach, Multiple Semantic Subspaces Network (MSSN), to simplify the complex and intractable semantic features. Our method generates multiple disentangled subfeatures via direct sum decomposition, which not only retain completely semantic information, but also obtain simple independent hierarchical features. Specially, instead of original space, using projection subspace to map embedding space can reduce the difficulty of model optimization and enhance the generalization ability of the model. Full experiments are achieved on almost all datasets and comparision with many algorithms which exist in the present latest literature. Compared with nineteen competitors, the results show that our model outperforms the state-of-the-art on conventional ZSL setting and has a very competitive performance on generalized ZSL setting. (C) 2020 Elsevier B.V. All rights reserved.

关键词： Zero shot learning Subspace Direct sum decomposition 68T05

来源：评论

学校读者我要写书评

暂无评论

D⁴M: dataset Distillation via Disentangled Diffusion Model

D<SUP>4</SUP>M: Dataset Distillation via Disentangled Diffus...

引用

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Su, Duo Hou, Junjie Gao, Weizhi Tian, Yingjie Tang, Bowen UCAS Sch Comp Sci & Technol Beijing Peoples R China UCAS Sino Danish Coll Beijing Peoples R China NCSU Dept Comp Sci Raleigh NC USA UCAS Sch Econ & Management Beijing Peoples R China Chinese Acad Sci Res Ctr Fictitious Econ & Data Sci Beijing Peoples R China Chinese Acad Sci Key Lab Big Data Min & Knowledge Management Beijing Peoples R China UCAS MOE Social Sci Lab Digital Econ Forecasts & Polic Beijing Peoples R China Chinese Acad Sci Inst Comp Technol Beijing Peoples R China

ISBN: (纸本)9798350353013;9798350353006

dataset distillation offers a lightweight synthetic dataset for fast network training with promising test accuracy. To imitate the performance of the original dataset, most approaches employ bi-level optimization and the distillation space relies on the matching architecture. Nevertheless, these approaches either suffer significant computational costs on large-scale datasets or experience performance decline on cross-architectures. We advocate for designan economical dataset distillation framework that is independent of the matching architectures. With empirical observations, we argue that constraining the consistency of the real and synthetic image spaces will enhance the cross-architecture generalization. Motivated by this, we introduce dataset Distillation via Disentangled Diffusion Model ((DM)-M-4), an efficient framework for dataset distillation. Compared to architecture-dependent methods, D4(M) employs latent diffusion model to guarantee consistency and incorporates label information into category prototypes. The distilled datasets are versatile, eliminating the need for repeated generation of distinct datasets for various architectures. Through comprehensive experiments, (DM)-M-4 demonstrates superior performance and robust generalization, surpassing the SOTA methods across most aspects.

关键词： Computer vision data-Centric AI dataset Distillation Latent Diffusion Model Prototype Learning

来源：评论

学校读者我要写书评

暂无评论

Rethinking label Flipping Attack: From Sample Masking to Sample Thresholding

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2023年第6期45卷 7668-7685页

作者： Xu, Qianqian Yang, Zhiyong Zhao, Yunrui Cao, Xiaochun Huang, Qingming Chinese Acad Sci Inst Comp Technol Key Lab Intelligent Informat Proc Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Comp Sci & Technol Beijing 100049 Peoples R China Sun Yat Sen Univ Sch Cyber Sci & Technol Shenzhen Campus Shenzhen 518107 Peoples R China Univ Chinese Acad Sci Sch Comp Sci & Technol Beijing 101408 Peoples R China Univ Chinese Acad Sci Key Lab Big Data Min & Knowledge Management BDKM Beijing 101408 Peoples R China Chinese Acad Sci Inst Comp Technol Key Lab Intelligent Informat Proc Beijing 100190 Peoples R China Peng Cheng Lab Shenzhen 518055 Peoples R China

Nowadays, machine learning (ML) and deep learning (DL) methods have become fundamental building blocks for a wide range of AI applications. The popularity of these methods also makes them widely exposed to malicious attacks, which may cause severe security concerns. To understand the security properties of the ML/DL methods, researchers have recently started to turn their focus to adversarial attack algorithms that could successfully corrupt the model or clean data owned by the victim with imperceptible perturbations. In this paper, we study the label Flipping Attack (LFA) problem, where the attacker expects to corrupt an ML/DL model's performance by flipping a small fraction of the labels in the training data. Prior art along this direction adopts combinatorial optimization problems, leading to limited scalability toward deep learning models. To this end, we propose a novel minimax problem which provides an efficient reformulation of the sample selection process in LFA. In the new optimization problem, the sample selection operation could be implemented with a single thresholding parameter. This leads to a novel training algorithm called Sample Thresholding. Since the objective function is differentiable and the model complexity does not depend on the sample size, we can apply Sample Thresholding to attack deep learning models. Moreover, since the victim's behavior is not predictable in a poisonous attack setting, we have to employ surrogate models to simulate the true model employed by the victim model. Seeing the problem, we provide a theoretical analysis of such a surrogate paradigm. Specifically, we show that the performance gap between the true model employed by the victim and the surrogate model is small under mild conditions. On top of this paradigm, we extend Sample Thresholding to the crowdsourced ranking task, where labels collected from the annotators are vulnerable to adversarial attacks. Finally, experimental analyses on three real-world datasets speak

关键词： data models Training data Training Deep learning Predictive models Testing Optimization label flipping attack machine learning

来源：评论

学校读者我要写书评

暂无评论

A Tale of HodgeRank and Spectral Method: Target Attack Against Rank Aggregation is the Fixed Point of Adversarial Game

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2023年第4期45卷 4090-4108页

作者： Ma, Ke Xu, Qianqian Zeng, Jinshan Li, Guorong Cao, Xiaochun Huang, Qingming Univ Chinese Acad Sci Sch Comp Sci & Technol Beijing 100049 Peoples R China Chinese Acad Sci Inst Comp Technol Key Lab Intelligent Informat Proc Beijing 100190 Peoples R China Jiangxi Normal Univ Sch Comp & Informat Engn Nanchang 330022 Jiangxi Peoples R China Sun Yat sen Univ Sch Cyber Sci & Technol Shenzhen Campus Shenzhen 518107 Guangdong Peoples R China Univ Chinese Acad Sci Key Lab Big Data Min & Knowledge Management Beijing 100049 Peoples R China Chinese Acad Sci Inst Comp Technol Key Lab Intelligent Informat Proc Beijing 100190 Peoples R China Peng Cheng Lab Shenzhen 518055 Guangdong Peoples R China

Rank aggregation with pairwise comparisons has shown promising results in elections, sports competitions, recommendations, and information retrieval. However, little attention has been paid to the security issue of such algorithms, in contrast to numerous research work on the computational and statistical characteristics. Driven by huge profit, the potential adversary has strong motivation and incentives to manipulate the ranking list. Meanwhile, the intrinsic vulnerability of the rank aggregation methods is not well studied in the literature. To fully understand the possible risks, we focus on the purposeful adversary who desires to designate the aggregated results by modifying the pairwise data in this paper. From the perspective of the dynamical system, the attack behavior with a target ranking list is a fixed point belonging to the composition of the adversary and the victim. To perform the targeted attack, we formulate the interaction between the adversary and the victim as a game-theoretic framework consisting of two continuous operators while Nash equilibrium is established. Then two procedures against HodgeRank and RankCentrality are constructed to produce the modification of the original data. Furthermore, we prove that the victims will produce the target ranking list once the adversary masters the complete information. It is noteworthy that the proposed methods allow the adversary only to hold incomplete information or imperfect feedback and perform the purposeful attack. The effectiveness of the suggested target attack strategies is demonstrated by a series of toy simulations and several real-world data experiments. These experimental results show that the proposed methods could achieve the attacker's goal in the sense that the leading candidate of the perturbed ranking list is the designated one by the adversary.

关键词： Voting Sports Electric potential Training Stochastic processes Security Optimization Adversarial learning pairwise comparison ranking aggregation

来源：评论

学校读者我要写书评

暂无评论

Impact of Consumer Characteristics on IP Product Purchase: Empirical Evidence from a Chinese Chain Retailer 10

Impact of Consumer Characteristics on IP Product Purchase: E...

引用

10th International Conference on Information Technology and Quantitative management, ITQM 2023

作者： Tang, Jiayu Wang, Tingyan Tian, Xin Cai, Qiang Research Center on Fictitious Economy and Data Science Chinese Academy of Sciences Beijing100190 China School of Economics and Management University of Chinese Academy of Sciences Beijing100190 China Key Laboratory of Big Data Mining and Knowledge Management Chinese Academy of Sciences Beijing100190 China Sino-Danish College University of Chinese Academy of Sciences Beijing100190 China Nuffield Department of Medicine University of Oxford South Parks Road OxfordOX1 3SY United Kingdom NIHR Oxford Biomedical Research Centre Big Data Institute University of Oxford Old Road Campus OxfordOX3 7LF United Kingdom School of Economics and Management Ningbo University of Technology Ningbo315211 China

Developing IP product has become a common differentiation strategy for retailers to attract customers and remain competence strength. This study uses empirical model to relate consumer characteristics with IP product purchase. The results show that consumers shopping on weekends are more likely to buy IP products, and consumers who have stocking behavior are less likely to purchase IP products. Consumers who are members of the retailer also have lower probability to buy IP products. These findings provide suggestions for the retailers who want to apply differentiation strategy by introducing co-branded IP products. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://***/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the Tenth International Conference on Information Technology and Quantitative management.

关键词： Consumer behavior

来源：评论

学校读者我要写书评

暂无评论

Deep Feature Inpainting for Unsupervised Visual Anomaly Detection 10

Deep Feature Inpainting for Unsupervised Visual Anomaly Dete...

引用

10th International Conference on Information Technology and Quantitative management, ITQM 2023

作者： Yang, Jie Lyu, Mengjin Qi, Zhiquan Tian, Yingjie Shi, Yong School of Mathematical Science University of Chinese Academy of Sciences Beijing101408 China School of Economics and Management University of Chinese Academy of Sciences Beijing101408 China Research Center on Fictitious Economy and Data Science Chinese Academy of Sciences Beijing100190 China Key Laboratory of Big Data Mining and Knowledge Management Chinese Academy of Sciences Beijing100190 China College of Information Science and Technology University of Nebraska OmahaNE68182 United States School of Computer Science and Technology University of Chinese Academy of Sciences Beijing101408 China

Visual anomaly detection is an important and challenging problem in the field of machine learning and computer vision. In this paper, we address the problem of unsupervised visual anomaly detection from the perspective of image feature restoration. We propose deep feature inpainting (DFI) where we develop a deep learning model that can automatically inpainting the normal image patterns or features from the anomalous images by eliminating the irregular structures or patterns. By comparing the anomalous features and their reconstructed counterparts, we can identify and analyze the anomalous structures in the images. Extensive experiments have been carried out to demonstrate the effectiveness of the approach. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://***/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the Tenth International Conference on Information Technology and Quantitative management.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

RLPS: A Reinforcement Learning-Based Framework for Personalized Search

引用

ACM TRANSACTIONS ON INFORMATION SYSTEMS 2021年第3期39卷 p1-29页

作者： Yao, Jing Dou, Zhicheng Xu, Jun Wen, Ji-Rong Renmin Univ China Sch Informat 59 Zhongguancun St Beijing 100872 Peoples R China Renmin Univ China Sch Artificial Intelligence 59 Zhongguancun St Beijing 100872 Peoples R China MOE Beijing Key Lab Big Data Management & Anal Method Key Lab Data Engn & Knowledge Engn 59 Zhongguancun St Beijing 100872 Peoples R China

Personalized search is a promising way to improve search qualities by taking user interests into consideration. Recently, machine learning and deep learning techniques have been successfully applied to search result personalization. Most existing models simply regard the personal search history as a static set of user behaviors and learn fixed ranking strategies based on all the recorded data. Though improvements have been achieved, the essence that the search process is a sequence of interactions between the search engine and user is ignored. The user's interests may dynamically change during the search process, therefore, it would be more helpful if a personalized search model could track the whole interaction process and adjust its ranking strategy continuously. In this article, we adapt reinforcement learning to personalized search and propose a framework, referred to as RLPS. It utilizes a Markov Decision Process (MDP) to track sequential interactions between the user and search engine, and continuously update the underlying personalized ranking model with the user's real-time feedback to learn the user's dynamic interests. Within this framework, we implement two models: the listwise RLPS-L and the hierarchical RLPS-H. RLPS-L interacts with users and trains the ranking model with document lists, while RLPS-H improves model training by designing a layered structure and introducing document pairs. In addition, we also design a feedback-aware personalized ranking component to capture the user's feedback, which impacts the user interest profile for the next query. Significant improvements over existing personalized search models are observed in the experiments on the public AOL search log and a commercial log.

关键词： Personalized search reinforcement learning Markov decision process (MDP)

来源：评论

学校读者我要写书评

暂无评论

Cryptanalysis of an optical cryptosystem with uncertainty quantification in a probabilistic model

引用

APPLIED OPTICS 2022年第19期61卷 5567-5574页

作者： Sun, Xinkai Zhang, Sanguo Shi, Yishi Univ Chinese Acad Sci Sch Math Sci Beijing 100049 Peoples R China Chinese Acad Sci Key Lab Big Data Min & Knowledge Management Beijing 100049 Peoples R China Pazhou Lab Guangzhou 510320 Peoples R China Univ Chinese Acad Sci Sch Optoelect Beijing 100049 Peoples R China Univ Chinese Acad Sci Ctr Mat Sci & Optoelect Engn Beijing 100049 Peoples R China

In this paper, a modified probabilistic deep learning method is proposed to attack the double random phase encryption by modeling the conditional distribution of plaintext. The well-trained probabilistic model gives both predictions of plaintext and uncertainty quantification, the latter of which is first introduced to optical cryptanalysis. Predictions of the model are close to real plaintexts, showing the success of the proposed model. Uncertainty quantification reveals the level of reliability of each pixel in the prediction of plaintext without ground truth. Subsequent simulation experiments demonstrate that uncertainty quantification can effectively identify poor-quality predictions to avoid the risk of unreliability from deep learning models. (C) 2022 Optica Publishing Group

关键词： Fourier transforms Ghost imaging Image quality Neural networks Optical encryption Optical systems

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Based Image Quality Assessment: A Survey 10

Deep Learning Based Image Quality Assessment: A Survey

引用

10th International Conference on Information Technology and Quantitative management, ITQM 2023

作者： Yang, Jie Lyu, Mengjin Qi, Zhiquan Shi, Yong School of Mathematical Science University of Chinese Academy of Sciences Beijing101408 China School of Economics and Management University of Chinese Academy of Sciences Beijing101408 China Research Center on Fictitious Economy and Data Science Chinese Academy of Sciences Beijing100190 China Key Laboratory of Big Data Mining and Knowledge Management Chinese Academy of Sciences Beijing100190 China College of Information Science and Technology University of Nebraska OmahaNE68182 United States School of Computer Science and Technology University of Chinese Academy of Sciences Beijing101408 China

Image quality assessment (IQA) is the problem of measuring the perceptual quality of images, which is crucial for many image-related applications. It is a difficult task due to the coupling of various degradation and the scarcity of annotations. To facilitate a better understanding of IQA, we survey the recent advances in deep learning based IQA methods, which have demonstrated remarkable performance and innovation in this field. We classify the IQA methods into two main groups: reference-based and reference-free methods. Reference-based methods compare query images with reference images, while reference-free methods do not. We further subdivide reference-based methods into full-reference and reduced-reference methods, depending on the amount of information they need from the reference images, and reference-free methods into single-input, pair-input, and multimodal-input methods, according to the form of input they use. The advantages and limitations of each category are analyzed and some representative examples of state-of-the-art methods are provided. We conclude our paper by highlighting some of the future directions and open challenges in deep learning based IQA. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://***/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the Tenth International Conference on Information Technology and Quantitative management.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Exploring Freight Loading management by Deep Learning: a Case Study in Home Furnishing Industry

引用

Annals of data Science 2022年第2期9卷 213-228页

作者： Deng, Wei Patil, Rajvardhan Liu, Fangyao Daji, Ergu Shi, Yong College of Big Data Statistics Guizhou University of Finance and Economics Guizhou550025 China College of Computer and Information Science Arkansas Tech University RussellvilleAR68182 United States College of Information Science and Technology University of Nebraska at Omaha OmahaNE68182 United States Research Center on Fictitious Economy and Data Science The Chinese Academy of Sciences Beijing100190 China Key Laboratory of Big Data Mining and Knowledge Management Chinese Academy of Sciences Beijing100190 China School of Economics and Management University of Chinese Academy of Sciences Beijing100190 China Nebraska Furniture Mart OmahaNE68124 United States Division of Science Research and Technology Southwest Minzu University Chengdu610041 China

Long short-term memory (LSTM) networks as state-of-the-art Deep Learning models, have achieved remarkable results in time series forecasting. However, they are less commonly applied to the industry of logistics. This paper presents two novel LSTM networks to predict the freight loading of routing areas, and the design of a smart loading management system is introduced. While most existing works on LSTM utilize the power of its prediction very well, our study shows less accurate and instable results if only LSTM network is applied, that arise from the big variation in the dataset. Instead, two constraints are inspired in this paper. The first constraint adds a time factor node to strengthen the correlation between predicted final loading units and booked loading units as close to delivery date. And the second proposal extends to parallel training by groupwise constraint. Experiments with across 3-year records in a national wide home furnishing retail company show that constraint LSTM networks significantly improve both the accuracy and stability of the prediction. Besides, the design of a smart loading management system will be shown, in which LSTM model plays the core rule to predict loading capacity, meanwhile the rule-based system triggers different events based on loading prediction and truck size. © 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

关键词： Long short-term memory

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：