检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Zhu, Tinghui Liu, Jingping Liang, Jiaqing Jiang, Haiyun Xiao, Yanghua Wang, Zongyu Xie, Rui Xian, Yunsen Shanghai Key Laboratory of Data Science School of Computer Science Fudan University Shanghai China School of Information Science and Engineering East China University of Science and Technology Shanghai China Meituan Shanghai China

Taxonomy expansion task is essential in organizing the ever-increasing volume of new concepts into existing taxonomies. Most existing methods focus exclusively on using textual semantics, leading to an inability to generalize to unseen terms and the "Prototypical Hypernym Problem." In this paper, we propose Visual Taxonomy Expansion (VTE), introducing visual features into the taxonomy expansion task. We propose a textual hypernymy learning task and a visual prototype learning task to cluster textual and visual semantics. In addition to the tasks on respective modalities, we introduce a hyper-proto constraint that integrates textual and visual semantics to produce fine-grained visual semantics. Our method is evaluated on two datasets, where we obtain compelling results. Specifically, on the Chinese taxonomy dataset, our method significantly improves accuracy by 8.75 %. Additionally, our approach performs better than ChatGPT on the Chinese taxonomy dataset. Copyright © 2023, The Authors. All rights reserved.

关键词： Taxonomies

来源：评论

学校读者我要写书评

暂无评论

Stabilize the latent space for image autoregressive modeling: a unified perspective 24

Stabilize the latent space for image autoregressive modeling...

引用

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

作者： Yongxin Zhu Bocheng Li Hang Zhang Xin Li Linli Xu Lidong Bing School of Data Science University of Science and Technology of China and State Key Laboratory of Cognitive Intelligence School of Computer Science and Technology University of Science and Technology of China and State Key Laboratory of Cognitive Intelligence Zhejiang University

ISBN: (纸本)9798331314385

Latent-based image generative models, such as Latent Diffusion Models (LDMs) and Mask Image Models (MIMs), have achieved notable success in image generation tasks. These models typically leverage reconstructive autoencoders like VQGAN or VAE to encode pixels into a more compact latent space and learn the data distribution in the latent space instead of directly from pixels. However, this practice raises a pertinent question: Is it truly the optimal choice? In response, we begin with an intriguing observation: despite sharing the same latent space, autoregressive models significantly lag behind LDMs and MIMs in image generation. This finding contrasts sharply with the field of NLP, where the autoregressive model GPT has established a commanding presence. To address this discrepancy, we introduce a unified perspective on the relationship between latent space and generative models, emphasizing the stability of latent space in image generative modeling. Furthermore, we propose a simple but effective discrete image tokenizer to stabilize the latent space for image generative modeling by applying K-Means on the latent features of self-supervised learning models. Experimental results show that image autoregressive modeling with our tokenizer (DiGIT) benefits both image understanding and image generation with the next token prediction principle, which is inherently straightforward for GPT models but challenging for other generative models. Remarkably, for the first time, a GPT-style autoregressive model for images outperforms LDMs, which also exhibits substantial improvement akin to GPT when scaling up model size. Our findings underscore the potential of an optimized latent space and the integration of discrete tokenization in advancing the capabilities of image generative models. The code is available at https://***/DAMO-NLP-SG/DiGIT.

关键词：

来源：评论

学校读者我要写书评

暂无评论

PE: A Poincare Explanation Method for Fast Text Hierarchy Generation

PE: A Poincare Explanation Method for Fast Text Hierarchy Ge...

引用

2024 Findings of the Association for Computational Linguistics, EMNLP 2024

作者： Chen, Qian Li, Dongyang He, Xiaofeng Li, Hongzhao Yi, Hongyu School of Computer Science and Technology East China Normal University Shanghai China NPPA Key Laboratory of Publishing Integration Development ECNUP Shanghai China Sichuan Caizi Software Information Network Co. Ltd. China

ISBN: (纸本)9798891761681

The black-box nature of deep learning models in NLP hinders their widespread *** research focus has shifted to Hierarchical Attribution (HA) for its ability to model feature *** works model non-contiguous combinations with a time-costly greedy search in Euclidean spaces, neglecting underlying linguistic information in feature *** this work, we introduce a novel method, namely Poincare Explanation (PE), for modeling feature interactions with hyperbolic spaces in a time efficient ***, we take building text hierarchies as finding spanning trees in hyperbolic *** we project the embeddings into hyperbolic spaces to elicit inherit semantic and syntax hierarchical *** we propose a simple yet effective strategy to calculate Shapley *** we build the the hierarchy with proving the constructing process in the projected space could be viewed as building a minimum spanning tree and introduce a time efficient building *** results demonstrate the effectiveness of our approach. © 2024 Association for Computational Linguistics.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Integrated Square Error of Hazard Rate Estimation for Survival data with Missing Censoring Indicators

引用

Journal of Systems science & Complexity 2021年第2期34卷 735-758页

作者： ZOU Yuye FAN Guoliang ZHANG Riquan School of Economics and Management Shanghai Maritime UniversityShanghai 201306China Key Laboratory of Advanced Theory and Application in Statistics and Data Science-MOE School of StatisticsEast China Normal UniversityShanghai 200062China

The problem of hazard rate estimation under right-censored assumption has been investigated *** square error(ISE)of estimation is one of the most widely accepted measurements of the global performance for nonparametric kernel *** there are no results available for ISE of hazard rate estimation under right-censored model with censoring indicators missing at random(MAR)so *** paper constructs an imputation estimator of the hazard rate function and establish asymptotic normality of the ISE for the kernel hazard rate estimator with censoring indicators *** the same time,an asymptotic representation of the mean integrated square error(MISE)is also *** finite sample behavior of the estimator is investigated via one simple simulation.

关键词： Asymptotic normality integrated square error missing at random right-censored model

来源：评论

学校读者我要写书评

暂无评论

Empowering Few-Shot Relation Extraction with The Integration of Traditional RE Methods and Large Language Models

arXiv

引用

arXiv 2024年

作者： Liu, Ye Zhang, Kai Gan, Aoran Yue, Linan Hu, Feng Liu, Qi Chen, Enhong School of Data Science School of Computer Science and Technology University of Science and Technology of China China State Key Laboratory of Cognitive Intelligence China

Few-Shot Relation Extraction (FSRE), a subtask of Relation Extraction (RE) that utilizes limited training instances, appeals to more researchers in Natural Language Processing (NLP) due to its capability to extract textual information in extremely low-resource scenarios. The primary methodologies employed for FSRE have been fine-tuning or prompt tuning techniques based on Pre-trained Language Models (PLMs). Recently, the emergence of Large Language Models (LLMs) has prompted numerous researchers to explore FSRE through In-Context Learning (ICL). However, there are substantial limitations associated with methods based on either traditional RE models or LLMs. Traditional RE models are hampered by a lack of necessary prior knowledge, while LLMs fall short in their task-specific capabilities for RE. To address these shortcomings, we propose a Dual-System Augmented Relation Extractor (DSARE), which synergistically combines traditional RE models with LLMs. Specifically, DSARE innovatively injects the prior knowledge of LLMs into traditional RE models, and conversely enhances LLMs' task-specific aptitude for RE through relation extraction augmentation. Moreover, an Integrated Prediction module is employed to jointly consider these two respective predictions and derive the final results. Extensive experiments demonstrate the efficacy of our proposed method. Copyright © 2024, The Authors. All rights reserved.

关键词： Extraction

来源：评论

学校读者我要写书评

暂无评论

Weight Perturbation as Defense against Adversarial Word Substitutions

Weight Perturbation as Defense against Adversarial Word Subs...

引用

2022 Findings of the Association for Computational Linguistics: EMNLP 2022

作者： Xu, Jianhan Li, Linyang Zhang, Jiping Zheng, Xiaoqing Chang, Kai-Wei Hsieh, Cho-Jui Huang, Xuanjing School of Computer Science Fudan University Shanghai China Shanghai Key Laboratory of Intelligent Information Processing China Department of Computer Science University of California Los Angeles United States

The existence and pervasiveness of textual adversarial examples have raised serious concerns to security-critical applications. Many methods have been developed to defend against adversarial attacks for neural natural language processing (NLP) models. Adversarial training is one of the most successful defense methods by adding some random or intentional perturbations to the original input texts and making the models robust to the perturbed examples. In this study, we explore the feasibility of improving the adversarial robustness of NLP models by performing perturbations in the parameter space rather than the input feature space. The weight perturbation helps to find a better solution (i.e., the values of weights) that minimizes the adversarial loss among other feasible solutions. We found that the weight perturbation can significantly improve the robustness of NLP models when it is combined with the perturbation in the input embedding space, yielding the highest accuracy on both clean and adversarial examples across different datasets. © 2022 Association for Computational Linguistics.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Analysis of Stock Splits Based on Risk Theory: Empirical Evidence from the Chinese Stock Markets

引用

Journal of Systems science and Information 2022年第1期10卷 19-34页

作者： Shujin WU Tong XU MOE Key Laboratory of Advanced Theory and Application in Statistics and Data Science School of StatisticsEast China Normal UniversityShanghai 200062China School of Statistics East China Normal UniversityShanghai 200062China

The paper first analyzes price change due to stock splits in Chinese stock markets,which shows stock prices typically go up for stock *** theoretical analyses based on risk theory are presented to explain the reason,where the method comes from a new perspective and obtained theoretical conclusions show that stock splits typically make stock price go up if risk-compensation function is convex,and go down if risk-compensation function is *** prices typically go up for stock splits because risk-compensation functions are mainly *** obtained conclusions are consistent with the known results in the last three decades.

关键词： stock split risk theory price Chinese stock markets

来源：评论

学校读者我要写书评

暂无评论

New regularization techniques for ill-conditioning problems and their applications: Choices of regularization parameters

引用

Engineering Analysis with Boundary Elements 2023年第1期152卷 347-361页

作者： Zhang, Li-Ping Li, Zi-Cai Wei, Yimin Huang, Hung-Tsai School of Mathematical Sciences Qufu Normal University Qufu Shandong273165 China Department of Applied Mathematics National Sun Yat-sen University Kaohsiung80424 Taiwan Shanghai Key Laboratory of Contemporary Mathematics and School of Mathematics Fudan University Shanghai200433 China Department of Data Science and Analytics I-Shou University Kaohsiung84001 Taiwan

This is a continued study of Li et al. (2011) for the regularization of ill-conditioning problems, where the minimal singular value σmin of discrete matrices is infinitesimal. To remove the effects of high frequency caused by the singular vector of σmin better, we combine the Tikhonov regularization (TR) with the truncated singular value decomposition (TSVD) (simply denoted by T-TR). New computational formulas of the traditional condition number (Cond) and the effective condition number (Cond_eff) are derived, and a brief error analysis is made. The regularization parameter λ is involved in the TR, and its better (or optimal) choices are essential in both theory and computation. For ill-conditioning problems, stability is more imperative than accuracy. The most important criterion is to reduce the Cond. Consider the linear algebraic equations Ax = b. The optimal regularization parameter for the TR is derived as λ=σmaxσmin, where σmax and σmin(>0) are the maximal and the minimal singular values of matrix A, respectively. The L-curve techniques emphasize the ill-conditioning of ‖xλ‖ (see Hansen (1998)). Note that the norm ‖xλ‖ is only part of the Cond in stability. In fact, by using regularization, the Cond may be greatly reduced, while the errors do not increase much. The second important criterion is for both stability and solution accuracy. The sensitivity index proposed in Zhang et al. (2021) indicates the severity of ill-conditioning via accuracy for comparing different numerical methods and techniques. The sensitivity index is effective not only to select the source nodes in the method of fundamental solutions (MFS), but also to select the regularization parameter in the TR. The regularization choices of the T-TR are similar to those of the TR because differences of the Cond and the errors between the TR and the T-TR are insignificant. Numerical experiments are reported by the MFS for Laplace's equation, to support the new regularization techniques proposed. For

关键词： Errors

来源：评论

学校读者我要写书评

暂无评论

Watch Out for Your Agents! Investigating Backdoor Threats to LLM-Based Agents 38

Watch Out for Your Agents! Investigating Backdoor Threats to...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Yang, Wenkai Bi, Xiaohan Lin, Yankai Chen, Sishuo Zhou, Jie Sun, Xu Gaoling School of Artificial Intelligence Renmin University of China Beijing China Center for Data Science Peking University China Pattern Recognition Center WeChat AI Tencent Inc. China National Key Laboratory for Multimedia Information Processing School of Computer Science Peking University China

Driven by the rapid development of Large Language Models (LLMs), LLM-based agents have been developed to handle various real-world applications, including finance, healthcare, and shopping, etc. It is crucial to ensure the reliability and security of LLM-based agents during applications. However, the safety issues of LLM-based agents are currently under-explored. In this work, we take the first step to investigate one of the typical safety threats, backdoor attack, to LLM-based agents. We first formulate a general framework of agent backdoor attacks, then we present a thorough analysis of different forms of agent backdoor attacks. Specifically, compared with traditional backdoor attacks on LLMs that are only able to manipulate the user inputs and model outputs, agent backdoor attacks exhibit more diverse and covert forms: (1) From the perspective of the final attacking outcomes, the agent backdoor attacker can not only choose to manipulate the final output distribution, but also introduce the malicious behavior in an intermediate reasoning step only, while keeping the final output correct. (2) Furthermore, the former category can be divided into two subcategories based on trigger locations, in which the backdoor trigger can either be hidden in the user query or appear in an intermediate observation returned by the external environment. We implement the above variations of agent backdoor attacks on two typical agent tasks including web shopping and tool utilization. Extensive experiments show that LLM-based agents suffer severely from backdoor attacks and such backdoor vulnerability cannot be easily mitigated by current textual backdoor defense algorithms. This indicates an urgent need for further research on the development of targeted defenses against backdoor attacks on LLM-based agents. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Incentivizing Crowdsensing for DT-Enabled Metaverse 20th

Incentivizing Crowdsensing for DT-Enabled Metaverse

引用

20th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2024

作者： Ye, Dongdong Huang, Xumin Wu, Yuan Kang, Jiawen Zhong, Weifeng Niyato, Dusit School of Automation Guangdong University of Technology Guangzhou China State Key Laboratory of Internet of Things for Smart City University of Macau Taipa China Department of Computer and Information Science University of Macau Taipa China Collge of Computing and Data Science Nanyang Technological University Singapore Singapore

ISBN: (纸本)9789819628292

Digital Twin (DT) has been proposed as a key enabling technology to produce a digital clone of the physical world and facilitate the convergence of the physical and virtual worlds for building a real-world Metaverse. For physical-virtual synchronization, crowdsensing is applied to employ Internet of Things (IoT) devices to sense physical objects and collect status information for timely updating the associated DTs. However, an incentive mechanism is necessitated to stimulate the devices to perform sensing tasks while meeting task requirements such as Age of Information (AoI) requirements. Thus, we adopt a contract theoretic approach to design an incentive mechanism for the devices that are classified into several types and the type parameters are unknown to others. The contract consists of a set of contract items with each contract item representing a device type, which specifies the required data updating frequency and monetary reward. We formulate an optimal contract design problem subject to feasible constraints including the average AoI requirement for each device type. Furthermore, we solve the problem by transforming it into a convex optimization problem that can be directly tackled by the existing methods. Finally, we provide numerical results to demonstrate the effectiveness and efficiency of our scheme. © IFIP International Federation for Information Processing 2025.

关键词： Convex optimization

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：