检索结果-内蒙古大学图书馆

Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and data Mining V.1

作者： Yubao Tang Ruqing Zhang Jiafeng Guo Maarten de Rijke Shihao Liu Shuaiqiang Wang Dawei Yin Xueqi Cheng CAS Key Lab of Network Data Science and Technology ICT CAS Beijing China University of the Chinese Academy of Sciences Beijing China & University of Amsterdam Amsterdam Netherlands CAS Key Lab of Network Data Science and Technology ICT CAS Beijing China & University of Chinese Academy of Sciences Beijing China University of Amsterdam Amsterdam Netherlands Baidu Inc. Beijing China

ISBN: (纸本)9798400712456

In book search, relevant book information should be returned in response to a query. Books contain complex, multi-faceted information such as metadata, outlines, and main text, where the outline provides hierarchical information between chapters and sections. Generative retrieval (GR) is a new retrieval paradigm that consolidates corpus information into a single model to generate identifiers of documents that are relevant to a given query. How can GR be applied to book search? Directly applying GR to book search is a challenge due to the unique characteristics of book search: The model needs to retain the complex, multi-faceted information of the book, which increases the demand for labeled data. Splitting book information and treating it as a collection of separate segments for learning might result in a loss of hierarchical *** propose an effective Generative retrieval framework for Book Search (GBS) that features two main components: data augmentation and outline-oriented book encoding. For data augmentation, GBS constructs multiple query-book pairs for training; it constructs multiple book identifiers based on the outline, various forms of book contents, and simulates real book retrieval scenarios with varied pseudo-queries. This includes coverage-promoting book identifier augmentation, allowing the model to learn to index effectively, and diversity-enhanced query augmentation, allowing the model to learn to retrieve effectively. Outline-oriented book encoding improves length extrapolation through bi-level positional encoding and retentive attention mechanisms to maintain context over long sequences. Experiments on a proprietary Baidu dataset demonstrate that GBS outperforms strong baselines, achieving a 9.8% improvement in terms of MRR@20, over the state-of-the-art RIPOR method. Experiments on public datasets confirm the robustness and generalizability of GBS, highlighting its potential to enhance book retrieval.

关键词： book retrieval

来源：评论

学校读者我要写书评

暂无评论

CLIPURE: PURIFICATION IN LATENT SPACE VIA CLIP FOR ADVERSARIALLY ROBUST ZERO-SHOT CLASSIFICATION

arXiv

引用

arXiv 2025年

作者： Zhang, Mingkun Bi, Keping Chen, Wei Guo, Jiafeng Cheng, Xueqi State Key Lab of AI Safety CAS Key Lab of AI Safety Institute of Computing Technology Chinese Academy of Sciences Beijing China CAS Key Laboratory of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

In this paper, we aim to build an adversarially robust zero-shot image classifier. We ground our work on CLIP, a vision-language pre-trained encoder model that can perform zero-shot classification by matching an image with text prompts "a photo of a .". Purification is the path we choose since it does not require adversarial training on specific attack types and thus can cope with any foreseen attacks. We then formulate purification risk as the KL divergence between the joint distributions of the purification process of denoising the adversarial samples and the attack process of adding perturbations to benign samples, through bidirectional Stochastic Differential Equations (SDEs). The final derived results inspire us to explore purification in the multi-modal latent space of CLIP. We propose two variants for our CLIPure approach: CLIPure-Diff which models the likelihood of images’ latent vectors with the DiffusionPrior module in DaLLE-2 (modeling the generation process of CLIP’s latent vectors), and CLIPure-Cos which models the likelihood with the cosine similarity between the embeddings of an image and "a photo of a.". As far as we know, CLIPure is the first purification method in multi-modal latent space and CLIPure-Cos is the first purification method that is not based on generative models, which substantially improves defense efficiency. We conducted extensive experiments on CIFAR-10, ImageNet, and 13 datasets that previous CLIP-based defense methods used for evaluating zero-shot classification robustness. Results show that CLIPure boosts the SOTA robustness by a large margin, e.g., from 71.7% to 91.1% on CIFAR10, from 59.6% to 72.6% on ImageNet, and 108% relative improvements of average robustness on the 13 datasets over previous SOTA. The code is available at https://***/TMLResearchGroup-cas/CLIPure. Copyright © 2025, The Authors. All rights reserved.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

Explainable and Layout-Aware Timing Prediction 24

Explainable and Layout-Aware Timing Prediction

引用

43rd International Conference on Computer-Aided Design, ICCAD 2024

作者： Lyu, Zhengyang Li, Xiaqing Du, Zidong Guo, Qi University of Science and Technology of China Hefei China State Key Lab of Processors Institute of Computing Techology CAS Beijing China Shanghai Innovation Center for Processor Technologies Shanghai China Ministry of Education Key Lab for Big Data and AI in Transportation Beijing Jiaotong University Beijing China

ISBN: (纸本)9798400710773

Accurate and fast timing prediction at early design stages is crucial for achieving timing closure in very-large-scale integration (VLSI) design. Machine learning (ML) based approaches have been widely adopted for timing prediction to pursue both of accuracy and speed. Unfortunately, these approaches fail to fully exploit the layout features which can significantly impact the timing metrics, thus leading to great accuracy degradation. In this paper, we propose LE-Timing, an end-to-end explainable and layout-aware timing prediction approach that can achieve better accuracy by leveraging important layout information. The key insight is that crucial information about coupling capacitance between routing wires can significantly impact timing metrics, and fully exploiting this layout information in the timing prediction model can substantially improve its accuracy. However, efficiently leveraging layout features while maintaining accuracy and interpretability poses three key challenges: node-level layout-graph integration, interpretability and training efficiency. To address these, we further propose three key technologies specially designed for LE-Timing correspondingly, including a layout-encoding net model, an explainable cell model, and an explicit-message-passing scheme. Experimental results on 14 open-source designs with advanced 7-nm technology node demonstrate that LE-Timing achieves R2 score 0.92 for arrival time prediction and represents strong interpretability. © 2024 Copyright is held by the owner/author(s).

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

NeuVSA: A Unified and Efficient Accelerator for Neural Vector Search

NeuVSA: A Unified and Efficient Accelerator for Neural Vecto...

引用

IEEE Symposium on High-Performance Computer Architecture

作者： Ziming Yuan Lei Dai Wen Li Jie Zhang Shengwen Liang Ying Wang Cheng Liu Huawei Li Xiaowei Li Jiafeng Guo Peng Wang Renhai Chen Gong Zhang State Key Lab of Processors Institute of Computing TechnologyCAS University of Chinese Academy of Sciences School of Computer and Information Technology Shanxi University School of Computer Science Peking University Zhongguancun National Laboratory Beijing Key Lab of Network Data Science and Technology Institute of Computing Technology CAS Huawei Technologies Co. Ltd. China

ISBN: (数字)9798331506476

ISBN: (纸本)9798331506483

Neural Vector Search (NVS) has exhibited superior search quality over traditional key-based strategies for information retrieval tasks. An effective NVS architecture requires high recall, low latency, and high throughput to enhance user experience and cost-efficiency. However, implementing NVS on existing neural network accelerators and vector search accelerators is sub-optimal due to the separation between the embedding stage and vector search stage at both algorithm and architecture levels. Fortunately, we unveil that Product Quantization (PQ) opens up an opportunity to break separation. However, existing PQ algorithms and accelerators still focus on either the embedding stage or the vector search stage, rather than both simultaneously. Simply combining existing solutions still follows the beaten track of separation and suffers from insufficient parallelization, frequent data access conflicts, and the absence of scheduling, thus failing to reach optimal recall, latency, and throughput. To this end, we propose a unified and efficient NVS accelerator dubbed NeuVSA based on algorithm and architecture co-design philosophy. Specifically, on the algorithm level, we propose a learned PQ-based unified NVS algorithm that consolidates two separate stages into the same computing and memory access paradigm. It integrates an end-to-end joint training strategy to learn the optimal codebook and index for enhanced recall and reduced PQ complexity, thus achieving smoother acceleration. On the architecture level, we customize a homogeneous NVS accelerator based on the unified NVS algorithm. Each sub-accelerator is optimized to exploit all parallelism exposed by unified NVS, incorporating a structured index assignment strategy and an elastic on-chip buffer to alleviate buffer conflicts for reduced latency. All sub-accelerators are coordinated using a hardware-aware scheduling strategy for boosted throughput. Experimental results show that the joint training strategy improves recall

关键词： Training Quantization (signal) Philosophical considerations Graphics processing units Computer architecture Throughput Vectors User experience System-on-chip Indexes

来源：评论

学校读者我要写书评

暂无评论

TrustRAG: An Information Assistant with Retrieval Augmented Generation

arXiv

引用

arXiv 2025年

作者： Fan, Yixing Yan, Qiang Wang, Wenshan Guo, Jiafeng Zhang, Ruqing Cheng, Xueqi Key Lab of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

Retrieval-Augmented Generation (RAG) has emerged as a crucial technique for enhancing large models with real-time and domain-specific knowledge. While numerous improvements and open-source tools have been proposed to refine the RAG framework for accuracy, relatively little attention has been given to improving the trustworthiness of generated results. To address this gap, we introduce TrustRAG, a novel framework that enhances RAG from three perspectives: indexing, retrieval, and generation. Specifically, in the indexing stage, we propose a semantic-enhanced chunking strategy that incorporates hierarchical indexing to supplement each chunk with contextual information, ensuring semantic completeness. In the retrieval stage, we introduce a utility-based filtering mechanism to identify high-quality information, supporting answer generation while reducing input length. In the generation stage, we propose fine-grained citation enhancement, which detects opinion-bearing sentences in responses and infers citation relationships at the sentence-level, thereby improving citation accuracy. We open-source the TrustRAG framework and provide a demonstration studio designed for excerpt-based question answering tasks 1. Based on these, we aim to help researchers: 1) systematically enhancing the trustworthiness of RAG systems and (2) developing their own RAG systems with more reliable outputs. © 2025, CC BY.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Adversarial training via multi-guidance and historical memory enhancement

引用

Neurocomputing 2025年 619卷

作者： Zhao, Chenyu Qian, Yaguan Wang, Bin Gu, Zhaoquan Ji, Shouling Wang, Wei Zhang, Yanchun School of Science Zhejiang University of Science and Technology China Network and Data Security China China School of Computer Science and Technology Zhejiang University China Education Key Lab for Intelligent Networks and Network Security Xi'an Jiaotong University China Institute for Sustainable Industries and Liveable Cities Victoria University Australia

Deep neural networks (DNNs) are often susceptible to the influence of adversarial examples, potentially leading to severe security issues. Adversarial training stands out as one of the most effective defenses. In this paper, we empirically demonstrate that weakly robust models maintain resilience to adversarial examples generated on a fully trained robust model. Building on this observation, we introduce a novel weight ensemble training method named Multi-Guided Adversarial Training (MGAT). MGAT improves both clean and adversarial accuracy through two guiding strategies: self-guidance and expert-guidance. Self-guidance promotes diversity in decision boundaries among member models, while expert-guidance aligns them more closely with natural decision boundaries. Additionally, MGAT enhances the historical memory of model weights using an exponential moving average (EMA) with a memory factor, thereby better memorizing the weights of weakly robust models. Our experiments demonstrate that MGAT is highly effective in defending against a range of adversarial attacks. Notably, MGAT achieves an accuracy of 55.48% against AutoAttack, marking a 10.92% improvement over standard adversarial training on CIFAR-10. © 2024 Elsevier B.V.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

A Model-Agnostic Hierarchical Framework Towards Trajectory Prediction

引用

Journal of Computer science and technology 2025年第2期40卷 322-339页

作者： Qian, Tang-Wen Wang, Yuan Xu, Yong-Jun Zhang, Zhao Wu, Lin Qiu, Qiang Wang, Fei Domain-Oriented Intelligent System Research Center Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China University of Chinese Academy of Sciences Beijing100190 China CAS Key Laboratory of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China

Predicting the future trajectories of multiple agents is essential for various applications in real life, such as surveillance systems, autonomous driving, and social robots. The trajectory prediction task is influenced by many factors, including the individual historical trajectory, interactions between agents, and the fuzzy nature of the observed agents’ motion. While existing methods have made great progress on the topic of trajectory prediction, they treat all the information uniformly, which limits the effectiveness of information utilization. To this end, in this paper, we propose and utilize a model-agnostic framework to regard all the information in a two-level hierarchical view. Particularly, the first-level view is the inter-trajectory view. In this level, we observe that the difficulty in predicting different trajectory samples varies. We define trajectory difficulty and train the proposed framework in an "easy-to-hard" schema. The second-level view is the intra-trajectory level. We find the influencing factors for a particular trajectory can be divided into two parts. The first part is global features, which keep stable within a trajectory, i.e., the expected destination. The second part is local features, which change over time, i.e., the current position. We believe that the two types of information should be handled in different ways. The hierarchical view is beneficial to take full advantage of the information in a fine-grained way. Experimental results validate the effectiveness of the proposed model-agnostic framework. © Institute of Computing technology, Chinese Academy of sciences 2025.

关键词： Hierarchical systems

来源：评论

学校读者我要写书评

暂无评论

Wi-SFDAGR: WiFi-Based Cross-Domain Gesture Recognition via Source-Free Domain Adaptation

引用

IEEE Internet of Things Journal 2025年

作者： Yan, Huan Zhang, Xiang Huang, Jinyang Feng, Yuanhao Li, Meng Wang, Anzhi Ou, Weihua Wang, Hongbing Liu, Zhi Guizhou Normal University School of Big Data and Computer Science Guiyang550025 China University of Science and Technology of China CAS Key Laboratory of Electromagnetic Space Information Hefei230026 China Hefei University of Technology School of Computer and Information Hefei230601 China The University of Electro-Communications Department of Computer and Network Engineering Tokyo1828585 Japan

WiFi Channel State Information (CSI)-based gesture recognition offers unique advantages, including cost-effectiveness and enhanced privacy protection, and has garnered significant attention in recent years. However, existing WiFi-based gesture recognition solutions exhibit poor generalization ability when deployed in new environment, orientation, or location. Although some methods combine labeled source domain and unlabeled target domain to learn domain-independent features, factors such as data privacy protection hinder access to source data during practical environment adaptation. Consequently, we consider realistic scenario where source data is unavailable during adaptation of unlabeled test data, and instead, a trained source domain model is used. In this paper, we propose Wi-SFDAGR, a WiFi-based Source-free Domain Adaptation Gesture Recognition framework. Specifically, we treat cross-domain as an unsupervised clustering problem, aiming to ensure that features within local neighborhoods exhibit similar prediction results while those farther apart display different prediction outcomes in the feature space. We theoretically analyze the effect of enhanced prediction consistency between neighbor points extracted from gestures on generalization error. Furthermore, we employ an attraction-dispersion network to strengthen prediction consistency among closely located features in the feature space while reducing it for distantly located features. To mitigate noise introduced during nearest neighbor sample selection in the feature space (where predictions may not align with the input sample’s prediction), we progressively improve nearby sample feature aggregation by estimating uncertainty to reweight local neighborhood predictions. Finally, extensive experiments are conducted on the Widar 3.0 and XRF55 datasets and the results show our proposed framework outperforms most cross-domain methods. © 2014 IEEE.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

Bridging Queries and Tables through Entities in Table Retrieval

arXiv

引用

arXiv 2025年

作者： Li, Da Bi, Keping Guo, Jiafeng Cheng, Xueqi CAS Key Lab of Network Data Science and Technology ICT CAS China University of Chinese Academy of Sciences Beijing China

Table retrieval is essential for accessing information stored in structured tabular formats;however, it remains less explored than text retrieval. The content of the table primarily consists of phrases and words, which include a large number of entities, such as time, locations, persons, and organizations. Entities are well-studied in the context of text retrieval, but there is a noticeable lack of research on their applications in table retrieval. In this work, we explore how to leverage entities in tables to improve retrieval performance. First, we investigate the important role of entities in table retrieval from a statistical perspective and propose an entity-enhanced training framework. Subsequently, we use the type of entities to highlight entities instead of introducing an external knowledge base. Moreover, we design an interaction paradigm based on entity representations. Our proposed framework is plug-and-play and flexible, making it easy to integrate into existing table retriever training processes. Empirical results on two table retrieval benchmarks, NQ-TABLES and OTT-QA, show that our proposed framework is both simple and effective in enhancing existing retrievers. We also conduct extensive analyses to confirm the efficacy of different components. Overall, our work provides a promising direction for elevating table retrieval, enlightening future research in this area. Copyright © 2025, The Authors. All rights reserved.

关键词： Knowledge engineering

来源：评论

学校读者我要写书评

暂无评论

Sensing-Communication-Computation Integration for Federated Edge Learning with Controllable Model Dropout

引用

IEEE Internet of Things Journal 2025年

作者： Jiao, Xiang Zhu, Guangxu Jiang, Wei Chen, Li Luo, Wu Wen, Dingzhu State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Peking University Beijing China Shenzhen Research Institute of Big Data Shenzhen China The State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Peking University Beijing China The CAS Key Laboratory of Wireless Optical Communication University of Science and Technology of China Hefei China Network Intelligence Center School of Information Science and Technology ShanghaiTech University Shanghai China

FEderated Edge Learning (FEEL) is an advanced paradigm in edge artificial intelligence, enabling privacy-preserving collaborative model training through periodic communication between edge devices and a central server. FEEL involves three key processes: sensing, computation, and communication for data acquisition, processing, and exchange, respectively. Due to limited system resources, optimizing each process individually may lead to suboptimal learning performance. This challenge has sparked research into Integrated Sensing-Computation-Communication (ISCC) design for enhanced FEEL. While previous work has optimized general learning parameters such as batch size and computing frequency, there is a lack of customized designs considering the neural network architecture as an optimizable variable in ISCC for FEEL. To close this gap, we introduce a novel design where each device generates a sub-model through controllable weight dropout, adding flexibility by directly manipulating the learning process and reducing computation and communication overhead. To guide ISCC resource allocation in this new setting, we present a comprehensive convergence analysis, revealing the tight coupling of sensing, computation, and communication across devices and their impact on FEEL convergence. Building on these theoretical insights, we formulate an ISCC problem aiming to maximize the FEEL convergence rate through joint optimization of variables such as batch size, sensing power, dropout rate, and communication power. This non-convex problem is decomposed into two sub-problems via alternating optimization: one controls batch size using a sorting algorithm, while the other focuses on ISCC device parameters, transformable into a convex problem solved by successive convex approximation. Extensive experiments using human motion recognition datasets demonstrate the superiority of the proposed design over baseline schemes. © 2014 IEEE.

关键词： Federated learning

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：