检索结果-内蒙古大学图书馆

32nd ACM International Conference on Multimedia, MM 2024

作者： Zhang, Ling Ma, Yidong Jiang, Zhi He, Weilei Bao, Zhongyun Fu, Gang Xu, Wenju Xiao, Chunxia School of Computer Science and Technology Hubei Key Laboratory of Intelligent Information Processing and Realtime Industrial Systems Wuhan University of Science and Technology Wuhan China School of Computer Science Wuhan University Wuhan China Department of Computing Hong Kong Polytechnic University Hong Kong Amazon Palo Alto United States

ISBN: (纸本)9798400706868

Recently, learning-based methods have made significant progress for image specular highlight removal. However, many of these approaches treat all the image pixels uniformly, overlooking the negative impact of invalid pixels on feature reconstruction. This oversight often leads to undesirable outcomes, such as color distortion or residual highlights. In this paper, we propose a novel image specular highlight removal network called HighlightRNet, which utilizes valid pixels as references to reconstruct the highlight-free image. To achieve this, we introduce a context-aware fusion block (CFBlock) that aggregates information in four directions, effectively capturing global contextual information. Additionally, we introduce a location-aware feature transformation module (LFTModule) to adaptively learn the valid pixels for feature reconstruction, thereby avoiding information errors caused by invalid pixels. With these modules, our method can produce high-quality highlight-free results without color distortion and highlight residual. Furthermore, we develop a multiple light image-capturing system to construct a large-scale highlight dataset called NSH, which exhibits minimal misalignment in image pairs and minimal brightness variation in non-highlight regions. Experimental results on various datasets demonstrate the superiority of our method over state-of-the-art methods, both qualitatively and quantitatively. © 2024 ACM.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

Improving Bert Fine-Tuning via Stabilizing Cross-Layer Mutual information 48

Improving Bert Fine-Tuning via Stabilizing Cross-Layer Mutua...

引用

48th IEEE International Conference on Acoustics, Speech and Signal processing, ICASSP 2023

作者： Li, Jicun Li, Xingjian Wang, Tianyang Wang, Shi Cao, Yanan Xu, Chengzhong Dou, Dejing Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China Big Data Lab Baidu Research China University of Alabama Birmingham United States Institute of Information Engineering Chinese Academy of Sciences China State Key Lab of Iotsc University of Macau China

ISBN: (纸本)9781728163277

Fine-tuning pre-trained language models, such as BERT, has shown enormous success among various NLP tasks. Though simple and effective, the process of fine-tuning has been found unstable, which often leads to unexpected poor performance. To increase stability and generalizability, most existing works resort to maintaining the parameters or representations of pre-trained models during fine-tuning. Nevertheless, very little work explores mining the reliable part of pre-learned information that can help to stabilize fine-tuning. To address this challenge, we introduce a novel solution in which we fine-tune BERT with stabilized cross-layer mutual information. Our method aims to preserve the reliable behaviors of cross-layer information propagation, instead of preserving the information itself, of the pre-trained model. Therefore, our method circumvents the domain conflicts between pre-trained and target tasks. We conduct extensive experiments with popular pre-trained BERT variants on NLP datasets, demonstrating the universal effectiveness and robustness of our method. © 2023 IEEE.

关键词： Fine-tuning Stability Mutual information Pre-trained Language Model

来源：评论

学校读者我要写书评

暂无评论

Exploring Query Efficient Data Generation towards Data-free Model Stealing in Hard Label Setting

arXiv

引用

arXiv 2024年

作者： Pei, Gaozheng Lyu, Shaojie Ma, Ke Yang, Pinci Xu, Qianqian Sun, Yingfei School of Electronic Electrical and Communication Engineering UCAS Beijing China Key Laboratory of Intelligent Information Processing Institute of Computing Technology CAS Beijing China Tencent Corporate

Data-free model stealing involves replicating the functionality of a target model into a substitute model without accessing the target model’s structure, parameters, or training data. The adversary can only access the target model’s predictions for generated samples. Once the substitute model closely approximates the behavior of the target model, attackers can exploit its white-box characteristics for subsequent malicious activities, such as adversarial attacks. Existing methods within cooperative game frameworks often produce samples with high confidence for the prediction of the substitute model, which makes it difficult for the substitute model to replicate the behavior of the target model. This paper presents a new data-free model stealing approach called Query Efficient Data Generation (QEDG). We introduce two distinct loss functions to ensure the generation of sufficient samples that closely and uniformly align with the target model’s decision boundary across multiple classes. Building on the limitation of current methods, which typically yield only one piece of supervised information per query, we propose the query-free sample augmentation that enables the acquisition of additional supervised information without increasing the number of queries. Motivated by theoretical analysis, we adopt the consistency rate metric, which more accurately evaluates the similarity between the substitute and target models. We conducted extensive experiments to verify the effectiveness of our proposed method, which achieved better performance with fewer queries compared to the state-of-the-art methods on the real MLaaS scenario and five datasets. Copyright © 2024, The Authors. All rights reserved.

关键词： Structured Query Language

来源：评论

学校读者我要写书评

暂无评论

Bi-Temporal Feature Relational Distillation for On-Board Lightweight Change Detection in Remote Sensing Imagery

引用

IEEE Transactions on Aerospace and Electronic Systems 2025年

作者： Wang, Guoqing Chen, He Li, Jie Wang, Jue Liu, Wenchao Chen, Liang Beijing Institute of Technology National Key Laboratory of Science and Technology on Space-Born Intelligent Information Processing Beijing100081 China Shanghai Aerospace Electronic Technology Institute Shanghai201109 China

Deep learning models for remote sensing change detection (CD) require significant resources, challenging real-time applications on spaceborne devices. Knowledge distillation (KD) technology is a potential solution that balances model size and CD performance. However, existing KD methods struggle to effectively transfer the teacher's ability to respond to changed landcovers, and also face challenges in assisting the student model in detecting changes in landcovers with subtle visual characteristics. To overcome these limitations, a bi-temporal feature relational distillation (FRD) framework is proposed. The FRD framework includes two designed distillation components: bi-temporal feature distance distillation (BFDD) and contrastive cluster representation distillation (CCRD). BFDD guides the student model to learn the relationship between bi-temporal feature categories from the teacher model and introduces the iteration-wise random selection and bilinear interpolation strategy to solve feature mismatch problem between the teacher and the student. CCRD describes the pixel-level and clusterlevel feature distributions through the designed contrastive cluster (CC) module, guiding the student model to align the change/nonchange features with the teacher model, so as to alleviate the confusion of change/non-change features. Extensive experiments and analyses have confirmed that the proposed FRD framework can develop lightweight models capable of achieving performance comparable to large models, thereby enhancing suitability for realtime CD. © 1965-2011 IEEE.

关键词： Change detection

来源：评论

学校读者我要写书评

暂无评论

Instance-Aware Diffusion Implicit Process for Box-Based Instance Segmentation 26

Instance-Aware Diffusion Implicit Process for Box-Based Inst...

引用

26th European Conference on Artificial Intelligence, ECAI 2023

作者： Ren, Hao Liu, Xingsong Huang, Junjian Wan, Ru Pu, Jian Lu, Hong Shanghai Key Laboratory of Intelligent Information Processing School of Computer Science Fudan University China Institute of Science and Technology for Brain-inspired Intelligence Fudan University China Mogo.ai Information and Technology Co. Ltd. Shanghai China

ISBN: (纸本)9781643684369

The diffusion model has demonstrated impressive performance in image generation, but its potential for discriminative tasks such as instance segmentation remains unexplored. In this paper, we propose an Instance-aware Diffusion Implicit Process (IDIP) framework for instance segmentation based on boxes. During training, IDIP diffuses ground-truth boxes across various time steps, extracting corresponding Region of Interest (RoI) features. Dynamic convolution is then used to predict boxes and categories for each RoI, and the mask head generates masks from these predictions. During inference, IDIP iteratively refines randomly generated boxes with the denoising diffusion implicit model, while the mask head derives final masks from RoIs based on the refined boxes. Our method surpasses existing approaches on the COCO benchmark, requiring fewer training steps and less memory resources due to its dynamic design and instance-aware characteristic. © 2023 The Authors.

关键词： Diffusion

来源：评论

学校读者我要写书评

暂无评论

A Cross-domain PAPR Reduction Method for OTFS Modulation

A Cross-domain PAPR Reduction Method for OTFS Modulation

引用

IEEE International Conference on Communications in China (ICCC)

作者： Dongkai Zhou Siqiang Wang Zhong Zheng Jing Guo Zesong Fei Weihua Yu School of Information and Electronics Beijing Institute of Technology Beijing China National Key Laboratory of Science and Technology on Space-Born Intelligent Information Processing

ISBN: (数字)9798350378412

ISBN: (纸本)9798350378429

In this work, a cross-domain PAPR reduction scheme is proposed, which is a combination of the phase rotation in the Delay-Doppler (DD) domain and the iterative clipping filtering (ICF) in the time domain. The former technique is distortion-less to the signals while incurring additional signaling overhead between the transceivers. The latter technique is more effective for PAPR reduction, but it can cause increased signal detection errors. Therefore, by leveraging the advantages of the two techniques, the proposed scheme can achieve improved PAPR reduction, given a trade-off between the signaling overhead and the bit error rate (BER). Specifically, the phase-rotation vectors optimization problem in the DD domain is formulated, and an iterative algorithm based on successive convex approximation (SCA) method and penalty convex-concave procedure (PCCP) is employed to solve this non-convex problem. Considering the phase quantization and signaling overhead in practical systems, a low-overhead group randomization algorithm is proposed to select the phase mapping in the grouped phase rotation stage, while the linear filtering is applied in the ICF stage, which can flexibly adjust the signaling overhead and the BER. Compared to the existing methods, numerical simulations justify the advantages of the proposed scheme, which further reduces 1-3 dB PAPR, while causing a lower BER.

关键词： Time-frequency analysis Filtering Simulation Bit error rate Peak to average power ratio Transceivers Iterative methods

来源：评论

学校读者我要写书评

暂无评论

ES3: Evolving Self-Supervised Learning of Robust Audio-Visual Speech Representations

ES3: Evolving Self-Supervised Learning of Robust Audio-Visua...

引用

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Yuanhang Zhang Shuang Yang Shiguang Shan Xilin Chen Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS) Institute of Computing Technology CAS Beijing China University of Chinese Academy of Sciences Beijing China

ISBN: (数字)9798350353006

ISBN: (纸本)9798350353013

We propose a novel strategy, ES 3 , for self-supervised learning of robust audio-visual speech representations from unlabeled talking face videos. While many recent approaches for this task primarily rely on guiding the learning process using the audio modality alone to capture information shared between audio and video, we reframe the problem as the acquisition of shared, unique (modality-specific) and synergistic speech information to address the inherent asymmetry between the modalities. Based on this formulation, we propose a novel “evolving” strategy that progressively builds joint audio-visual speech representations that are strong for both uni-modal (audio & visual) and bi-modal (audio-visual) speech. First, we leverage the more easily learnable audio modality to initialize audio and visual representations by capturing audio-unique and shared speech information. Next, we incorporate video-unique speech information and bootstrap the audio-visual repre-sentations on top of the previously acquired shared knowledge. Finally, we maximize the total audio-visual speech information, including synergistic information to obtain robust and comprehensive representations. We implement ES 3 as a simple Siamese framework and experiments on both English benchmarks and a newly contributed large-scale Mandarin dataset show its effectiveness. In particular, on LRS2-BBC, our smallest model is on par with SoTA models with only 1/2 parameters and 1/8 unlabeled data (223h).

关键词： Visualization Computer vision Self-supervised learning Manuals Benchmark testing Data models Pattern recognition

来源：评论

学校读者我要写书评

暂无评论

Multi-View Consistent 3D GAN Inversion via Bidirectional Encoder

Multi-View Consistent 3D GAN Inversion via Bidirectional Enc...

引用

International Conference on Automatic Face and Gesture Recognition

作者： Haozhan Wu Hu Han Shiguang Shan Xilin Chen Key Laboratory of Intelligent Information Processing Chinese Academy of Sciences (CAS) Institute of Computing Technology CAS Beijing China University of the Chinese Academy of Sciences Beijing China

ISBN: (数字)9798350394948

ISBN: (纸本)9798350394955

3D GAN inversion enables not only 3D reconstruction from a 2D image, but also novel view synthesis and image editing. Existing works ensure the novel view synthesis quality by constraining the synthesized views to conform to the real image distribution. However, most of the methods did not consider the multi-view consistency, i.e., different photos of the same 3D scene via 3D GAN inversion can be inverted to the same 3D scene. In this paper, we propose a bidirectional encoder (BiDiE) for 3D GAN inversion that can improve the multi-view consistency and alleviate the interference of camera parameter prediction errors. On the one hand, the bidirectional encoder takes real images as input, estimates the camera parameters, and performs 3D reconstruction. On the other hand, the bidirectional encoder takes randomly sampled latent code and camera parameters as input, and generates synthesized images to assist in the latent code learning process. In addition, we extend the latent space from W+ to W++ to improve its reconstruction and editing capabilities. Experiments on the FFHQ, CelebA-HQ and Multi-PIE datasets prove that our proposed method outperforms state-of-the-art methods in multi-view consistent reconstruction as well as editing capability. 1 1 Code and datasets are available at https://***/WHZMM/BiDiE

关键词： Three-dimensional displays Codes Face recognition Interference Gesture recognition Cameras Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Disentangled latent factors for muti-cause treatment effect estimation 6

Disentangled latent factors for muti-cause treatment effect ...

引用

2024 6th International Conference on Artificial Intelligence Technologies and Applications, ICAITA 2024

作者： Zhao, Zhuoyue Wang, Haotian Xu, Liyang Wang, Qian Tang, Yuhua Department of Intelligent Data Science National University of Defense Technology Changsha410073 China Intelligent Game and Decision Lab. Academy of Military Science Beijing100071 China Institute for Quantum Information State Key Laboratory of High-Performance Computing College of Computer Science and Technology National University of Defense Technology Changsha410073 China

Existing methods estimate treatment effects from observational data and assume that covariates are all confounders. However, observed covariates may not directly represent confounding variables that influence both treatment and outcome. They always include variables that only affect the treatment or the outcome. In addition, for multi-dimensional binary treatments, disentangled methods are mainly designed for binary variables and ignore the impact of multi-cause treatment variables on the inference of latent factors. To address these two issues, based on variable decomposition and proxy inference, we propose the Disentangled Latent Factors for Multi-cause Treatment Estimation (DEMTE) algorithm. It utilizes an identifiable autoencoder to infer and disentangle latent factors based on the joint distribution of variables in observational data. DEMTE evaluates the treatment effect on the disentangled factors. Synthetic experiments and semi-synthetic experiments demonstrate the effectiveness of the inference and disentanglement techniques and our method achieves more accurate treatment effect estimation. © 2024 institute of Physics Publishing. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Unifying Sum Rate Analysis and Optimization of RIS-Assisted MU-MIMO Networks: An Operator-Valued Free Probability Approach

Unifying Sum Rate Analysis and Optimization of RIS-Assisted ...

引用

IEEE International Conference on Communications in China (ICCC)

作者： Siqiang Wang Zhong Zheng Jing Guo Zesong Fei School of Information and Electronics Beijing Institute of Technology Beijing China National Key Laboratory of Science and Technology on Space-Born Intelligent Information Processing

ISBN: (数字)9798350378412

ISBN: (纸本)9798350378429

The fundamental performance limit of wireless systems is closely related to the transmission strategy of the transceivers as well as the wireless channels. As the channels are determined by the intrinsic characteristics of the environments, traditional system design is transceiver-centric, which adapts the transmit signals to the environments. Sharply different from the transceiver-centric design, by introducing the reconfigurable intelligent surface (RIS) into wireless systems, both the transceivers and the channels can be designed by properly adjusting the deployment of RIS panels as well as the phase-change and amplification of the RIS elements. However, as the RIS induces structural changes to the wireless channels, it is difficult to analyze and optimize the RIS-assisted systems in a unified framework. In this paper, we adopt the operatorvalued free probability theory to analyze the ergodic sum rate of an uplink MU-MIMO communication systems assisted by multiple cascaded RIS panels, where the sum rate is derived for rather generic system configurations, allowing arbitrary number of transceivers’ antennas, number of RIS panels, and a generic fading model. The obtained sum rate expression also enables a unified optimization framework that jointly designs the multi-antenna transceivers and the RIS configurations. Finally, numerical results are provided to verify the accuracy of the derived ergodic sum rate, and to show the performance of the joint optimization algorithms.

关键词： Wireless communication Fading channels Accuracy Reconfigurable intelligent surfaces Approximation algorithms Transceivers Numerical models

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：