检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Gao, Haowen Pang, Liang Xu, Shicheng Qu, Leigang Chua, Tat-Seng Shen, Huawei Cheng, Xueqi CAS Key Laboratory of AI Safety Institute of Computing Technology Chinese Academy of Sciences China Sea-NExT Joint Lab National University of Singapore Singapore CAS Key Laboratory of AI Security Institute of Computing Technology Chinese Academy of Sciences China

With the rapid development of ai-generated content (aiGC), the creation of high-quality ai-generated videos has become faster and easier, resulting in the Internet being flooded with all kinds of video content. However, the impact of these videos on the content ecosystem remains largely unexplored. Video information retrieval remains a fundamental approach for accessing video content. Building on the observation that retrieval models often favor ai-generated content in ad-hoc and image retrieval tasks, we investigate whether similar biases emerge in the context of challenging video retrieval, where temporal and visual factors may further influence model behavior. To explore this, we first construct a comprehensive benchmark dataset containing both real and ai-generated videos, along with a set of fair and rigorous metrics to assess bias. This benchmark consists of 13,000 videos generated by two state-of-the-art open-source video generation models. We meticulously design a suite of rigorous metrics to accurately measure this preference, accounting for potential biases arising from the limited frame rate and suboptimal quality of aiGC videos. We then applied three off-the-shelf video retrieval models to perform retrieval tasks on this hybrid dataset. Our findings reveal a clear preference for ai-generated videos in retrieval. Further investigation shows that incorporating ai-generated videos into the training set of retrieval models exacerbates this bias. Unlike the preference observed in image modalities, we find that video retrieval bias arises from both unseen visual and temporal information, making the root causes of video bias a complex interplay of these two factors. To mitigate this bias, we fine-tune the retrieval models using a contrastive learning approach. The results of this study highlight the potential implications of ai-generated videos on retrieval systems and offer valuable insights for future research in this area. Our dataset and code are publicly a

关键词： Video analysis

来源：评论

学校读者我要写书评

暂无评论

The Mirage of Model Editing: Revisiting Evaluation in the Wild

arXiv

引用

arXiv 2025年

作者： Yang, Wanli Sun, Fei Tan, Jiajun Ma, Xinyu Cao, Qi Yin, Dawei Shen, Huawei Cheng, Xueqi CAS Key Laboratory of AI Safety Institute of Computing Technology CAS China University of Chinese Academy of Sciences China Baidu Inc. China

Despite near-perfect results in artificial evaluations, the effectiveness of model editing in real-world applications remains unexplored. To bridge this gap, we propose to study model editing in question answering (QA) by establishing a rigorous evaluation practice to assess the effectiveness of editing methods in correcting LLMs’ errors. It consists of QAEdit, a new benchmark derived from popular QA datasets, and a standardized evaluation framework. Our single editing experiments indicate that current editing methods perform substantially worse than previously reported (38.5% vs. ∼96%). Through module analysis and controlled experiments, we demonstrate that this performance decline stems from issues in evaluation practices of prior editing research. One key issue is the inappropriate use of teacher forcing in testing prevents error propagation by feeding ground truth tokens (inaccessible in real-world scenarios) as input. Furthermore, we simulate real-world deployment by sequential editing, revealing that current approaches fail drastically with only 1000 edits. Our analysis provides a fundamental reexamination of both the real-world applicability of existing model editing methods and their evaluation practices, and establishes a rigorous evaluation framework with key insights to advance reliable and practical model editing research. Copyright © 2025, The Authors. All rights reserved.

关键词：

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

Precise Integral in NeRFs: Overcoming the Approximation Errors of Numerical Quadrature

Precise Integral in NeRFs: Overcoming the Approximation Erro...

引用

IEEE Workshop on Applications of Computer Vision (WACV)

作者： Boyuan Zhang Zhenliang He Meina Kan Shiguang Shan Key Lab of AI Safety Institute of Computing Technology CAS China University of Chinese Academy of Sciences China

ISBN: (数字)9798331510831

ISBN: (纸本)9798331510848

Neural Radiance Fields (NeRFs) use neural networks to translate spatial coordinates to corresponding volume density and directional radiance, enabling realistic novel view synthesis through volume rendering. Rendering new viewpoints involves computing volume rendering integrals along rays, usually approximated by numerical quadrature because of lacking closed-form solutions. In this paper, utilizing Taylor expansion, we demonstrate that numerical quadrature causes inevitable approximation error in NeRF integrals due to ignoring the parameter associated with the Lagrange remainder. To mitigate the approximation error, we propose a novel neural field with segment representation as input to implicitly model the remainder parameter. In theory, our proposed method is proven to possess the potential to achieve fully precise rendering integral, as demonstrated by comprehensive experiments on several commonly used datasets with state-of-the-art results.

关键词： Training Geometry Computer vision Closed-form solutions Neural networks Neural radiance field Approximation error Rendering (computer graphics) Taylor series Numerical models

来源：评论

学校读者我要写书评

暂无评论

MIGE: A Unified Framework for Multimodal Instruction-Based Image Generation and Editing

arXiv

引用

arXiv 2025年

作者： Tian, Xueyun Li, Wei Xu, Bingbing Yuan, Yige Wang, Yuanzhuo Shen, Huawei CAS Key Laboratory of AI Safety Institute of Computing Technology Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

Despite significant progress in diffusion-based image generation, subject-driven generation and instruction-based editing remain challenging. Existing methods typically treat them separately, struggling with limited high-quality data and poor generalization. However, both tasks require capturing complex visual variations while maintaining consistency between inputs and outputs. Therefore, we propose MIGE, a unified framework that standardizes task representations using multimodal instructions. It treats subject-driven generation as creation on a blank canvas and instruction-based editing as modification of an existing image, establishing a shared input-output formulation. MIGE introduces a novel multimodal encoder that maps free-form multimodal instructions into a unified vision-language space, integrating visual and semantic features through a feature fusion mechanism. This unification enables joint training of both tasks, providing two key advantages: (1) Cross-Task Enhancement: By leveraging shared visual and semantic representations, joint training improves instruction adherence and visual consistency in both subject-driven generation and instruction-based editing. (2) Generalization: Learning in a unified format facilitates cross-task knowledge transfer, enabling MIGE to generalize to novel compositional tasks, including instruction-based subject-driven editing. Experiments show that MIGE excels in both subject-driven generation and instruction-based editing while setting a state-of-the-art in the new task of instruction-based subject-driven editing. Code and model have been publicly available at this link. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Synthetic View Augmentation for Sign Language Recognition 25

Synthetic View Augmentation for Sign Language Recognition

引用

Companion Proceedings of the ACM on Web Conference 2025

作者： Yuting Peng Peiqi Jiao Honggang Zou Yuecong Min Xilin Chen State Key Laboratory of AI Safety Institute of Computing Technology Chinese Academy of Sciences Beijing China and University of Chinese Academy of Sciences Beijing China State Key Laboratory of AI Safety Institute of Computing Technology Chinese Academy of Sciences Beijing China

ISBN: (纸本)9798400713316

Sign language, as a visual language, is inherently spatial and dynamic, characterized by various hand shapes, movements, facial expressions, and body postures. In real-world scenarios, sign language recognition (SLR) systems are deployed in diverse environments with varying lighting conditions, backgrounds, and user positions, and models trained on data from a single viewpoint often struggle to generalize effectively to new perspectives. To address these limitations, this paper focuses on cross-view SLR and proposes a synthetic view augmentation pipeline to facilitate SLR models learning viewpoint-invariant features during training, thereby improving their performance on unseen viewpoints. Additionally, we introduce the XV-SLR model, which incorporates multi-modality information further to enhance the robustness of cross-view sign language recognition. The proposed method achieves a top-1 accuracy of 56.87% under the RGB setting and 57.97% under the RGB-D setting and outperforms other approaches in the Cross-View Isolated Sign Language Recognition Challenge (CV-ISLR). Code is available at https://***/VIPL-SLP/XV-SLR.

关键词： cross-view recognition

来源：评论

学校读者我要写书评

暂无评论

Agent-SiMT: Agent-Assisted Simultaneous Translation With Large Language Models

IEEE Transactions on Audio, Speech and Language Processing

引用

IEEE Transactions on Audio, Speech and Language Processing 2025年 33卷 2074-2083页

作者： Shoutao Guo Shaolei Zhang Zhengrui Ma Min Zhang Yang Feng Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences (ICT/CAS) Beijing China University of Chinese Academy of Sciences Beijing China School of Computer Science and Technology Soochow University Suzhou China Key Laboratory of AI Safety Chinese Academy of Sciences Beijing China

Simultaneous Machine Translation (SiMT) generates target translations in real-time while reading the source sentence. It relies on a policy to determine the optimal timing for producing translations, aiming to achieve both low response latency and high translation quality. Existing SiMT methods typically utilize the traditional Transformer architecture, employing complicated dynamic programming algorithms to jointly optimize policy decisions and translation generation. While these methods excel at determining policies, their translation capabilities are suboptimal. In contrast, Large Language Models (LLMs), trained on extensive corpora, exhibit exceptional generation capabilities but struggle to learn translation policies through traditional Transformer-based training methods. To address these limitations, we introduce Agent-SiMT, a novel framework combining the strengths of LLMs and traditional SiMT methods. Agent-SiMT contains a policy-decision agent and a translation agent. The policy-decision agent is managed by a SiMT model, which determines the policy using partial source sentences and translations. The translation agent, powered by an LLM, generates the target translation according to the policy decisions. These two agents work collaboratively to perform SiMT. Experiments demonstrate that our method attains state-of-the-art performance.

关键词： Translation Hidden Markov models Transformers Training Large language models Collaboration Vocabulary Speech processing Memory modules Machine translation

来源：评论

学校读者我要写书评

暂无评论

Revisiting Robust RAG: Do We Still Need Complex Robust Training in the Era of Powerful LLMs?

arXiv

引用

arXiv 2025年

作者： Ding, Hanxing Tao, Shuchang Pang, Liang Wei, Zihao Chen, Liwei Xu, Kun Shen, Huawei Cheng, Xueqi Key Laboratory of AI Safety Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China Kuaishou Technology China

Retrieval-augmented generation (RAG) systems often suffer from performance degradation when encountering noisy or irrelevant documents, driving researchers to develop sophisticated training strategies to enhance their robustness against such retrieval noise. However, as large language models (LLMs) continue to advance, the necessity of these complex training methods is increasingly questioned. In this paper, we systematically investigate whether complex robust training strategies remain necessary as model capacity grows. Through comprehensive experiments spanning multiple model architectures and parameter scales, we evaluate various document selection methods and adversarial training techniques across diverse datasets. Our extensive experiments consistently demonstrate that as models become more powerful, the performance gains brought by complex robust training methods drop off dramatically. We delve into the rationale and find that more powerful models inherently exhibit superior confidence calibration, better generalization across datasets (even when trained with randomly selected documents), and optimal attention mechanisms learned with simpler strategies. Our findings suggest that RAG systems can benefit from simpler architectures and training strategies as models become more powerful, enabling more scalable applications with minimal complexity. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Following the Autoregressive Nature of LLM Embeddings via Compression and Alignment

arXiv

引用

arXiv 2025年

作者： Deng, Jingcheng Jiang, Zhongtao Pang, Liang Chen, Liwei Xu, Kun Wei, Zihao Shen, Huawei Cheng, Xueqi Key Laboratory of AI Safety Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China Kuaishou Technology China

A new trend uses LLMs as dense text encoders via contrastive learning. However, since LLM embeddings predict the probability distribution of the next token, they are inherently generative and distributive, conflicting with contrastive learning, which requires embeddings to capture full-text semantics and align via cosine similarity. This discrepancy hinders the full utilization of LLMs’ pre-training capabilities, resulting in inefficient learning. In response to this issue, we propose AutoRegEmbed, a new contrastive learning method built on embedding conditional probability distributions, which integrates two core tasks: information compression and conditional distribution alignment. The information compression task encodes text into the embedding space, ensuring that the embedding vectors capture global semantics. The conditional distribution alignment task focuses on aligning text embeddings with positive samples embeddings by leveraging the conditional distribution of embeddings while simultaneously reducing the likelihood of generating negative samples from text embeddings, thereby achieving embedding alignment and uniformity. Experimental results demonstrate that our method significantly outperforms traditional contrastive learning approaches and achieves performance comparable to state-of-the-art models when using the same amount of data. Our code is available at https://***/TrustedLLM/AutoRegEmbed Copyright © 2025, The Authors. All rights reserved.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

PERPLEXITY-TRAP: PLM-BASED RETRIEVERS OVERRATE LOW PERPLEXITY DOCUMENTS

arXiv

引用

arXiv 2025年

作者： Wang, Haoyu Dai, Sunhao Zhao, Haiyuan Pang, Liang Zhang, Xiao Wang, Gang Dong, Zhenhua Xu, Jun Wen, Ji-Rong Gaoling School of Artificial Intelligence Renmin University of China Beijing China CAS Key Laboratory of AI Safety Institute of Computing Technology Beijing China Huawei Noah’s Ark Lab Shenzhen China

Previous studies have found that PLM-based retrieval models exhibit a preference for LLM-generated content, assigning higher relevance scores to these documents even when their semantic quality is comparable to human-written ones. This phenomenon, known as source bias, threatens the sustainable development of the information access ecosystem. However, the underlying causes of source bias remain unexplored. In this paper, we explain the process of information retrieval with a causal graph and discover that PLM-based retrievers learn perplexity features for relevance estimation, causing source bias by ranking the documents with low perplexity higher. Theoretical analysis further reveals that the phenomenon stems from the positive correlation between the gradients of the loss functions in language modeling task and retrieval task. Based on the analysis, a causal-inspired inference-time debiasing method is proposed, called Causal Diagnosis and Correction (CDC). CDC first diagnoses the bias effect of the perplexity and then separates the bias effect from the overall estimated relevance score. Experimental results across three domains demonstrate the superior debiasing effectiveness of CDC, emphasizing the validity of our proposed explanatory framework 1 © 2025, CC BY.

关键词： Semantics

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：