检索结果-内蒙古大学图书馆

Proceedings of the 38th International Conference on Neural information processing Systems

作者： Yinxuan Huang Chengmin Gao Bin Li Xiangyang Xue Shanghai Key Laboratory of Intelligent Information Processing School of Computer Science Fudan University

ISBN: (纸本)9798331314385

Given the complexities inherent in visual scenes, such as object occlusion, a comprehensive understanding often requires observation from multiple viewpoints. Existing multi-viewpoint object-centric learning methods typically employ random or sequential viewpoint selection strategies. While applicable across various scenes, these strategies may not always be ideal, as certain scenes could benefit more from specific viewpoints. To address this limitation, we propose a novel active viewpoint selection strategy. This strategy predicts images from unknown viewpoints based on information from observation images for each scene. It then compares the object-centric representations extracted from both viewpoints and selects the unknown viewpoint with the largest disparity, indicating the greatest gain in information, as the next observation viewpoint. Through experiments on various datasets, we demonstrate the effectiveness of our active viewpoint selection strategy, significantly enhancing segmentation and reconstruction performance compared to random viewpoint selection. Moreover, our method can accurately predict images from unknown viewpoints.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Reversible Data Hiding for 3D Mesh Model Based on Block Modulus Encryption and Multi-MSB Prediction 10th

Reversible Data Hiding for 3D Mesh Model Based on Block Modu...

引用

10th International Conference of Pioneering computer Scientists, Engineers and Educators, ICPCSEE 2024

作者： Fu, Zihao Gao, Yang Lyu, Wanli Key Lab of Intelligent Computing and Signal Processing of Ministry of Education School of Computer Science and Technology Anhui University Hefei230601 China

ISBN: (纸本)9789819787456

Reversible data hiding in encrypted domain (RDH-ED) fortifies data security and privacy safeguards while upholding the original data’s integrity and accessibility. Current research on RDH-ED focuses on 2D images, while research on 3D mesh models is still immature. This paper introduces an RDH-ED method using block modulus encryption and multi-MSB prediction. Initially, the original mesh model is partitioned into non-overlapping subblocks of equal size, and then the vertices in each subblock are encrypted with the same key for additive modulus encryption, ensuring that the spatial correlation present in the original mesh subblocks remains preserved within the encrypted subblocks. Subsequently, the subblocks are disrupted one by one using the 3D Arnold Transform to enhance security. The vertices in each embeddable subblock are divided into a reference vertex and several embeddable vertices, where the multi-MSB prediction strategy is employed to allocate embedding room for each embeddable vertex. Finally, the secret information is embedded into the allocated room. Since the proposed method almost completely preserves the spatial correlation within each subblock, the achieved embedding rate surpasses that of all previous outstanding methods that rely on vacating room after encryption (VRAE). The experimental findings demonstrate that the proposed approach achieves an average embedding rate of 45.55 bits per vertex (bpv), surpassing the state-of-the-art method that achieves 25.63 bpv. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Mesh generation

来源：评论

学校读者我要写书评

暂无评论

Masking Cascaded Self-attentions for Few-Shot Font-Generation Transformer 17th

Masking Cascaded Self-attentions for Few-Shot Font-Generati...

引用

17th Asian Conference on computer Vision, ACCV 2024

作者： Ma, Jing Xiang, Xiang He, Yan National Key Lab of Multi-Spectral Information Intelligent Processing Technology School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan China

ISBN: (纸本)9789819609161

Few-shot Font Generation (FFG) is a practical technology widely used in designing artistic characters, handwriting imitation, and identification, etc., which aims to generate realistic font images with a few reference samples. Typically, convolutional neural networks (CNNs) are employed to learn the representations of style and content from the font images. However, owing to the locality of convolutional operations, CNNs are not good at capturing the global structure of fonts, which leads to the generated images with blurry components and distorted space layouts. To address this problem, we consider cascading self-attention modules to exploit long-range dependencies for font generation and propose a transformer-based approach called FGTr. Following the style-content disentanglement paradigm, FGTr contains two different transformer encoders to extract the style and content sequences. A multi-layer transformer decoder is adopted to merge the two sequences and generate target images. In order to smoothen the transition of patch edges, we utilize a Local Self-Attention Mask (LSAM) to restrict the attention scope of each patch to a fixed-size sliding window, which plugs into the Transformer with no extra parameters. We also propose an Auxiliary Generation Module (AGM) in favor of generating glyphs closer to the real. Extensive experiments demonstrate the effectiveness and superiority of our method compared with state-of-the-art CNN-based models. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Exploring Trial-and-Error in Deep Learning:Initial Application to Isotope Detection in Mass Spectrometry

引用

Big Data Mining and Analytics 2024年第4期7卷 1251-1261页

作者： Qihong Jiao Yuxiao Wang Yongshuai Wang Shiwei Sun Xuefeng Cui School of Computer Science and Technology Shandong UniversityQingdao 266237China Key Lab of Intelligent Information Processing Institute of Computing TechnologyChinese Academy of SciencesBeijing 100190China

Mass spectrometry plays a crucial role in biomedicine by detecting isotopes,contributing significantly to research,diagnostics,and therapy *** study introduces IsoFusion,a deep learning model designed to address isotope detection in raw mass *** than directly applying convolutional layers to all signal and noise peaks,IsoFusion employs a trial-and-error ***,it investigates all potential charge states(trials)and collects signal peaks around expected m/z values for each ***,convolutional layers extract features from each trial,which are fused to identify the correct ***,the reparameterization trick predicts isotope features based on this correct trial.A key advantage of IsoFusion is shared model parameters across all trials,enhancing feature learning for less common charge states using data from prevalent *** results show a significant accuracy improvement for charge state 5,reaching 99.42%,compared to DeepIso’s 43.36%.Moreover,IsoFusion achieves a 97.33%detection accuracy for isotopes,with 2.4%of detected isotopes previously unidentified by four commonly used methods.

关键词： Liquid Chromatography-Mass Spectrometry(LC-MS) isotope detection retention time prediction charge state prediction deep learning

来源：评论

学校读者我要写书评

暂无评论

Explainable Synthetic Image Detection through Diffusion Timestep Ensembling

arXiv

引用

arXiv 2025年

作者： Wu, Yixin Zhang, Feiran Shi, Tianyuan Yin, Ruicheng Wang, Zhenghua Gan, Zhenliang Wang, Xiaohua Lv, Changze Zheng, Xiaoqing Huang, Xuanjing School of Computer Science Fudan University Shanghai Key Laboratory of Intelligent Information Processing Shanghai China

Recent advances in diffusion models have enabled the creation of deceptively real images, posing significant security risks when misused. In this study, we reveal that natural and synthetic images exhibit distinct differences in the high-frequency domains of their Fourier power spectra after undergoing iterative noise perturbations through an inverse multi-step denoising process, suggesting that such noise can provide additional discriminative information for identifying synthetic images. Based on this observation, we propose a novel detection method that amplifies these differences by progressively adding noise to the original images across multiple timesteps, and train an ensemble of classifiers on these noised images. To enhance human comprehension, we introduce an explanation generation and refinement module to identify flaws located in AI-generated images. Additionally, we construct two new datasets, GenHard and GenExplain, derived from the GenImage benchmark, providing detection samples of greater difficulty and high-quality rationales for fake images. Extensive experiments show that our method achieves state-of-the-art performance with 98.91% and 95.89% detection accuracy on regular and harder samples, increasing a minimal of 2.51% and 3.46% compared to baselines. Furthermore, our method also generalizes effectively to images generated by other diffusion models. Our code and datasets will be made publicly available. Copyright © 2025, The Authors. All rights reserved.

关键词： Inverse problems

来源：评论

学校读者我要写书评

暂无评论

HARNESSING JOINT RAIN-/DETAIL-AWARE REPRESENTATIONS TO ELIMINATE INTRICATE RAINS

arXiv

引用

arXiv 2024年

作者： Ran, Wu Ma, Peirong He, Zhiquan Ren, Hao Lu, Hong School of Computer Science Fudan University China Shanghai Key Lab of Intelligent Information Processing China

Recent advances in image deraining have focused on training powerful models on mixed multiple datasets comprising diverse rain types and backgrounds. However, this approach tends to overlook the inherent differences among rainy images, leading to suboptimal results. To overcome this limitation, we focus on addressing various rainy images by delving into meaningful representations that encapsulate both the rain and background components. Leveraging these representations as instructive guidance, we put forth a Context-based Instance-level Modulation (CoI-M) mechanism adept at efficiently modulating CNN- or Transformer-based models. Furthermore, we devise a rain-/detail-aware contrastive learning strategy to help extract joint rain-/detail-aware representations. By integrating CoI-M with the rain-/detail-aware Contrastive learning, we develop CoIC1, an innovative and potent algorithm tailored for training models on mixed datasets. Moreover, CoIC offers insight into modeling relationships of datasets, quantitatively assessing the impact of rain and details on restoration, and unveiling distinct behaviors of models given diverse inputs. Extensive experiments validate the efficacy of CoIC in boosting the deraining ability of CNN and Transformer models. CoIC also enhances the deraining prowess remarkably when real-world dataset is included. Copyright © 2024, The Authors. All rights reserved.

关键词： Rain

来源：评论

学校读者我要写书评

暂无评论

ToolEyes: Fine-Grained Evaluation for Tool Learning Capabilities of Large Language Models in Real-world Scenarios 31

ToolEyes: Fine-Grained Evaluation for Tool Learning Capabili...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Ye, Junjie Li, Guanyu Gao, Songyang Huang, Caishuang Wu, Yilong Li, Sixian Fan, Xiaoran Dou, Shihan Ji, Tao Zhang, Qi Gui, Tao Huang, Xuanjing School of Computer Science Fudan University China Institute of Modern Languages and Linguistics Fudan University China Research Institute of Intelligent Complex Systems Fudan University China Shanghai Key Laboratory of Intelligent Information Processing China Pengcheng Laboratory China

ISBN: (纸本)9798891761964

Existing evaluations of tool learning primarily focus on validating the alignment of selected tools (e.g., various APIs) for large language models (LLMs) with expected outcomes. However, these approaches rely on a limited set of scenarios where answers can be pre-determined. Furthermore, a sole emphasis on outcomes disregards the complex capabilities required for LLMs to effectively use tools. To tackle this issue, we propose ToolEyes, a fine-grained system tailored for the evaluation of the LLMs' tool learning capabilities in authentic scenarios. The system meticulously examines seven real-world scenarios, analyzing five dimensions crucial to LLMs in tool learning: format alignment, intent comprehension, behavior planning, tool selection, and answer organization. Additionally, ToolEyes incorporates a tool library boasting approximately 600 tools, serving as an intermediary between LLMs and the physical world. Evaluations involving ten LLMs across three categories reveal a preference for specific scenarios and limited cognitive abilities in tool learning. Intriguingly, expanding the model size even exacerbates the hindrance to tool learning. The code and data are available at https://***/Junjie-Ye/ToolEyes. © 2025 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Embedding expert demonstrations into clustering buffer for effective deep reinforcement learning

引用

Frontiers of information Technology & Electronic Engineering 2023年第11期24卷 1541-1556页

作者： Shihmin WANG Binqi ZHAO Zhengfeng ZHANG Junping ZHANG Jian PU Shanghai Key Laboratory of Intelligent Information Processing School of Computer ScienceFudan UniversityShanghai 200433China Institute of Science and Technology for Brain-inspired Intelligence Fudan UniversityShanghai 200433China

As one of the most fundamental topics in reinforcement learning(RL),sample efficiency is essential to the deployment of deep RL *** most existing exploration methods that sample an action from different types of posterior distributions,we focus on the policy sampling process and propose an efficient selective sampling approach to improve sample efficiency by modeling the internal hierarchy of the ***,we first employ clustering methods in the policy sampling process to generate an action candidate *** we introduce a clustering buffer for modeling the internal hierarchy,which consists of on-policy data,off-policy data,and expert data to evaluate actions from the clusters in the action candidate set in the exploration *** this way,our approach is able to take advantage of the supervision information in the expert demonstration *** on six different continuous locomotion environments demonstrate superior reinforcement learning performance and faster convergence of selective *** particular,on the LGSVL task,our method can reduce the number of convergence steps by 46.7%and the convergence time by 28.5%.Furthermore,our code is open-source for *** code is available at https://***/Shihwin/SelectiveSampling.

关键词： Reinforcement learning Sample eficiency Sampling process Clustering methods Autonomous driving

来源：评论

学校读者我要写书评

暂无评论

CAMG: Context-Aware Moment Graph Network for Multimodal Temporal Activity Localization via Language 12th

CAMG: Context-Aware Moment Graph Network for Multimodal Tem...

引用

12th National CCF Conference on Natural Language processing and Chinese Computing, NLPCC 2023

作者： Hu, Yuelin Xu, Yuanwu Zhang, Yuejie Feng, Rui Zhang, Tao Lu, Xuequan Gao, Shang School of Computer Science Shanghai Key Laboratory of Intelligent Information Processing Shanghai Collaborative Innovation Center of Intelligent Visual Computing Fudan University Shanghai200433 China School of Information Management and Engineering Shanghai Key Laboratory of Financial Information Technology Shanghai University of Finance and Economics Shanghai200433 China School of Information Technology Deakin University Waurn PondsVIC3216 Australia

ISBN: (纸本)9783031446924

Temporal Activity Localization via Language (TALL) is a challenging task for language based video understanding, especially when a video contains multiple moments of interest and the language query has words describing complex context dependencies between the moments. Latest studies have proposed various ways to exploit the temporal context of adjacent moments, but two apparent limitations remained. First, only limited context information was encoded based on RNNs or 2-D convolutions, which highly depended on the pre-sorting of proposals and lacked flexibility. Second, semantically correlated content in different moments was ignored, i.e., semantic context. To address these limitations, we propose a novel GCN-based framework, i.e., Context-Aware Moment Graph (CAMG) network, to jointly model the temporal context and semantic context. Also, we design a multi-step fusion scheme to aggregate object, motion and textual features. A Query-Gated Integration Module is further designed to select queried objects and filter out noisy ones. Our model achieves superior performance to state-of-the-art methods on two widely-used benchmark datasets. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

UMIT: Unifying Medical Imaging Tasks via Vision-Language Models

arXiv

引用

arXiv 2025年

作者： Yu, Haiyang Yi, Siyang Niu, Ke Zhuo, Minghan Li, Bin Shanghai Key Laboratory of Intelligent Information Processing School of Computer Science Fudan University China

With the rapid advancement of deep learning, particularly in the field of medical image analysis, an increasing number of Vision-Language Models (VLMs) are being widely applied to solve complex health and biomedical challenges. However, existing research has primarily focused on specific tasks or single modalities, which limits their applicability and generalization across diverse medical scenarios. To address this challenge, we propose UMIT, a unified multi-modal, multi-task VLM designed specifically for medical imaging tasks. UMIT is able to solve various tasks, including visual question answering, disease detection, and medical report generation. In addition, it is applicable to multiple imaging modalities (e.g., X-ray, CT and PET), covering a wide range of applications from basic diagnostics to complex lesion analysis. Moreover, UMIT supports both English and Chinese, expanding its applicability globally and ensuring accessibility to healthcare services in different linguistic contexts. To enhance the model’s adaptability and task-handling capability, we design a unique two-stage training strategy and fine-tune UMIT with designed instruction templates. Through extensive empirical evaluation, UMIT outperforms previous methods in five tasks across multiple datasets. The performance of UMIT indicates that it can significantly enhance diagnostic accuracy and workflow efficiency, thus providing effective solutions for medical imaging applications. The code and adopted datasets are available at https://***/dzosamu/UMIT/tree/main. © 2025, CC0.

关键词： Visual languages

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：