检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Tian, Chunwei Zheng, Menghua Jiao, Tiancai Zuo, Wangmeng Zhang, Yanning Lin, Chia-Wen the School of Software Northwestern Polytechnical University Shaanxi Xi’an710129 China the National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology Shaanxi Xi’an710129 China the School of Computer Science and Technology Harbin Institute of Technology Heilongjiang Harbin150001 China the Peng Cheng Laboratory Laboratory Guangdong Shenzhen518055 China the School of Computer Science Northwestern Polytechnical University the National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology Shaanxi Xi’an710129 China the Department of Electrical Engineering the Institute of Communications Engineering National Tsing Hua University Hsinchu300 Taiwan

Popular convolutional neural networks mainly use paired images in a supervised way for image watermark removal. However, watermarked images do not have reference images in the real world, which results in poor robustness of image watermark removal techniques. In this paper, we propose a self-supervised convolutional neural network (CNN) in image watermark removal (SWCNN). SWCNN uses a self-supervised way to construct reference watermarked images rather than given paired training samples, according to watermark distribution. A heterogeneous U-Net architecture is used to extract more complementary structural information via simple components for image watermark removal. Taking into account texture information, a mixed loss is exploited to improve visual effects of image watermark removal. Besides, a watermark dataset is conducted. Experimental results show that the proposed SWCNN is superior to popular CNNs in image watermark removal. Codes can be obtained at https://***/hellloxiaotian/SWCNN. Copyright © 2024, The Authors. All rights reserved.

关键词： Textures

来源：评论

学校读者我要写书评

暂无评论

Why Does Little Robustness Help? A Further Step Towards Understanding Adversarial Transferability

arXiv

引用

arXiv 2023年

作者： Zhang, Yechao Hu, Shengshan Zhang, Leo Yu Shi, Junyu Li, Minghui Liu, Xiaogeng Wan, Wei Jin, Hai School of Cyber Science and Engineering Huazhong University of Science and Technology China School of Software Engineering Huazhong University of Science and Technology China School of Information and Communication Technology Griffith University Australia School of Computer Science and Technology Huazhong University of Science and Technology China National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Hubei Key Laboratory of Distributed System Security China Hubei Engineering Research Center on Big Data Security China Cluster and Grid Computing Lab

Adversarial examples for deep neural networks (DNNs) have been shown to be transferable: examples that successfully fool one white-box surrogate model can also deceive other black-box models with different architectures. Although a bunch of empirical studies have provided guidance on generating highly transferable adversarial examples, many of these findings fail to be well explained and even lead to confusing or inconsistent advice for practical use. In this paper, we take a further step towards understanding adversarial transferability, with a particular focus on surrogate aspects. Starting from the intriguing "little robustness" phenomenon, where models adversarially trained with mildly perturbed adversarial samples can serve as better surrogates for transfer attacks, we attribute it to a trade-off between two dominant factors: model smoothness and gradient similarity. Our research focuses on their joint effects on transferability, rather than demonstrating the separate relationships alone. Through a combination of theoretical and empirical analyses, we hypothesize that the data distribution shift induced by off-manifold samples in adversarial training is the reason that impairs gradient similarity. Building on these insights, we further explore the impacts of prevalent data augmentation and gradient regularization on transferability and analyze how the trade-off manifest in various training methods, thus building a comprehensive blueprint for the regulation mechanisms behind transferability. Finally, we provide a general route for constructing superior surrogates to boost transferability, which optimizes both model smoothness and gradient similarity simultaneously, e.g., the combination of input gradient regularization and sharpness-aware minimization (SAM), validated by extensive experiments. In summary, we call for attention to the united impacts of these two factors for launching effective transfer attacks, rather than optimizing one while ignoring the other,

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Researching of the Mongolian word encoding method based on Transformer Mongolian speech recognition 21

Researching of the Mongolian word encoding method based on T...

引用

21st Chinese National Conference on Computational Linguistic, CCL 2022

作者： Xiaoxu, Zhang Zhiqiang, Ma Zhiqiang, Liu Caijilahu, Bao Inner Mongolia University of Technology Huhhot010000 China Inner Mongolia Autonomous Region Software Service Engineering Technology Research Center Based on Big Data Huhhot010000 China

In view of the fact that the Transformer model cannot learn the correspondence between Mongolian words with control symbols and the speech in the Mongolian speech recognition task, which causes the model to not adapt to the Mongolian language. A Mongolian word encoding method for Transformer model is proposed. The method uses Mongolian letter features and word features for mixed encoding. By combining Mongolian letter information, the Transformer model can distinguish Mongolian words with control symbols, and learn Mongolian words and pronunciation. Correspondence between. On the IMUT-MC dataset, the Transformer model is con-structed and the ablation and comparison experiments of word feature encoding methods are carried out. The results of ablation exper-iments show that the word feature encoding method reduces HWER, WER, and SER by 23.4%, 6.9%, and 2.6%, respectively;the comparative experimental results show that the word feature encoding method is ahead of all methods, with HWER and WER reaching 11.8%, 19.8%. © 2022 China National Conference on Computational Linguistics Published under Creative Commons Attribution 4.0 International License.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Attention based Mongolian Speaker Feature Extraction 21

Attention based Mongolian Speaker Feature Extraction

引用

21st Chinese National Conference on Computational Linguistic, CCL 2022

作者： Fangyuan, Zhu Zhiqiang, Ma Zhiqiang, Liu Caijilahu, Bao Hongbin, Wang Inner Mongolia University of Technology Huhhot010000 China Inner Mongolia Autonomous Region Software Service Engineering Technology Research Center Based on Big Data Huhhot010000 China

The speaker features extracted by the speaker feature extraction model have low discrimination, which makes the Mongolian acoustic model unable to learn the discrimination information, resulting in the model unable to adapt to different speakers. A speaker adaptation method based on attention is proposed. The method introduces neural Turing machine for adaptation, adds a memory module to store the speaker features, uses the attention mechanism to calculate the similarity weight matrix between the speaker features in the memory module and the current voice speaker features, and recombines the weight matrix into the speaker features s-vector, so as to improve the discrimination between the speaker features. On the IMUT-MCT dataset, the ablation experiment, model adaptation experiment and case analysis of speaker feature extraction method are carried out. The experimental results show that comparing s-vector, i-vector and d-vector with different speaker characteristics, s-vector reduces SER and WER by 4.96% and 1.08% respective-ly compared with the other two methods;By comparing different Mongolian acoustic models, the performance of the proposed method is improved compared with the baseline. © 2022 China National Conference on Computational Linguistics Published under Creative Commons Attribution 4.0 International License.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Rethinking WMMSE: Can Its Complexity Scale Linearly With the Number of BS Antennas?

arXiv

引用

arXiv 2022年

作者： Zhao, Xiaotong Lu, Siyuan Shi, Qingjiang Luo, Zhi-Quan The School of Software Engineering Tongji University Shanghai201804 China Shenzhen Research Institute of Big Data Shenzhen518172 China The Chinese University of Hong Kong Shenzhen518172 China

Precoding design for maximizing weighted sum-rate (WSR) is a fundamental problem for downlink of massive multi-user multiple-input multiple-output (MU-MIMO) systems. It is well-known that this problem is generally NP-hard due to the presence of multi-user interference. The weighted minimum mean-square error (WMMSE) algorithm is a popular approach for WSR maximization. However, its computational complexity is cubic in the number of base station (BS) antennas, which is unaffordable when the BS is equipped with a large antenna array. In this paper, we consider the WSR maximization problem with either a sum-power constraint (SPC) or per-antenna power constraints (PAPCs). For the former, we prove that any nontrivial stationary point must have a low-dimensional subspace structure, and then propose a reduced-WMMSE (R-WMMSE) with linear complexity by exploiting the solution structure. For the latter, we propose a linear-complexity WMMSE approach, named PAPC-WMMSE, by using a novel recursive design of the algorithm. Both R-WMMSE and PAPC-WMMSE have simple closed-form updates and guaranteed convergence to stationary points. Simulation results verify the efficacy of the proposed designs, especially the much lower complexity as compared to the state-of-the-art approaches for massive MU-MIMO systems. Copyright © 2022, The Authors. All rights reserved.

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

A Topology Independent Fast Rerouting Scheme with Hop-by-Hop Forwarding Can Achieve a Full Protection

SSRN

引用

SSRN 2022年

作者： Geng, Haijun Meng, Zhuo Zhang, Han Shi, Xingang Wang, Zhiliang Yin, Xia Chen, Meng Guo, Yingya The School of Automation and Software Engineering Shanxi University School of Computer and Information Technology China The Institute for Network Sciences and Cyberspace Tsinghua University China The Department of Computer Science and Technology Tsinghua University China The School of Big Data and Software Engineering Chongqing University China The College of Computer and Data Science Fuzhou University China

To support critical applications or guarantee a high quality of service, Internet Service Providers (ISPs) generally deploy a technique called routing protection algorithm for local fast rerouting. Alternate next hops are stored alongside with the default next hops in a router’s forwarding table, and can be immediately activated to invoke a loop free repair path in face of link or node failures, or used for multipath transmission. However, the existing routing protection methods either cannot protect all possible single-failure scenarios in the network or needs the assistance of an auxiliary address. In response to the above problems, the present study proposes a single-failure routing protection algorithm based on hop by hop routing (SPA). The algorithm has the following three advantages: (1) supports hop-by-hop forwarding, (2) supports incremental deployment, and (3) can protect all possible singlefailure scenarios in the network as long as network keep connected. Extensive simulations show that, compared to the state-of-the-art schemes which are adopted by the industry, SPA not only can handle arbitrary single-failure situations in the network but also has a small path stretch. These advantages make SPA a good candidate for practical solution for IP fast reroute when network failure occurs. © 2022, The Authors. All rights reserved.

关键词： Internet service providers

来源：评论

学校读者我要写书评

暂无评论

A Uniform Representation Model for Oct-Based Fingerprint Presentation Attack Detection and Reconstruction

SSRN

引用

SSRN 2023年

作者： Zhang, Wentian Liu, Haozhe Liu, Feng Ramachandra, Raghavendra The National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Shenzhen518060 China College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China SZU Branch Shenzhen Institute of Artificial Intelligence and Robotics for Society China Gjøvik2818 Norway

The technology of optical coherence tomography (OCT) to fingerprint imaging opens up a new research potential for fingerprint recognition owing to its ability to capture depth information of the skin layers. Developing robust and high security Automated Fingerprint Recognition Systems (AFRSs) are possible if the depth information can be fully utilized. However, in existing studies, Presentation Attack Detection (PAD) and subsurface fingerprint reconstruction based on depth information are treated as two independent branches, resulting in high computation and complexity of AFRS building. Thus, this paper proposes a uniform representation model for OCT-based fingerprint PAD and subsurface fingerprint reconstruction. Firstly, we design a novel semantic segmentation network which only trained by real finger slices of OCT-based fingerprints to extract multiple subsurface structures from those slices (also known as B-scans). The latent codes derived from the network are directly used to effectively detect the PA since they contain abundant subsurface biological information, which is independent with PA materials and has strong robustness for unknown PAs. Meanwhile, the segmented subsurface structures are adopted to reconstruct multiple subsurface 2D fingerprints. Recognition can be easily achieved by using existing mature technologies based on traditional 2D fingerprints. Extensive experiments are carried on our own established database, which is the largest public OCT-based fingerprint database with 2449 volumes. In PAD task, our method can improve 0.33% Acc from the state-of-the-art method. For reconstruction performance, our method achieves the best performance with 0.834 mean Intersection of Union (mIOU) and 0.937 Pixel Accuracy (PA). By comparing with the recognition performance on surface 2D fingerprints (e.g. commercial and high resolution), the effectiveness of our proposed method on high quality subsurface fingerprint reconstruction is further proved. © 2023, The

关键词： Optical tomography

来源：评论

学校读者我要写书评

暂无评论

Deconfound Semantic Shift and Incompleteness in Incremental Few-shot Semantic Segmentation 39

Deconfound Semantic Shift and Incompleteness in Incremental ...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Wu, Yirui Xia, Yuhang Li, Hao Yuan, Lixin Chen, Junyang Liu, Jun Lu, Tong Wan, Shaohua Key Laboratory of Water Big Data Technology of Ministry of Water Resources Hohai University Nanjing China College of Computer Science and Software Engineering Shenzhen University Shenzhen China School of Computing and Communication Lancaster University Lancaster United Kingdom National Key Lab for Novel Software Technology Nanjing University Nanjing China Shenzhen Institute for Advanced Study University of Electronic Science and Technology of China Shenzhen China

ISBN: (纸本)157735897X

Incremental few-shot semantic segmentation (IFSS) expands segmentation capacity of the trained model to segment new-class images with few samples. However, semantic meanings may shift from background to object class or vice versa during incremental learning. Moreover, new-class samples often lack representative attribute features when the new class greatly differs from the pre-learned old class. In this paper, we propose a causal framework to discuss the cause of semantic shift and incompleteness in IFSS, and we deconfound the revealed causal effects from two aspects. First, we propose a Causal Intervention Module (CIM) to resist semantic shift. CIM progressively and adaptively updates prototypes of old class, and removes the confounder in an intervention manner. Second, a Prototype Refinement Module (PRM) is proposed to complete the missing semantics. In PRM, knowledge gained from the episode learning scheme assists in fusing features of new-class and old-class prototypes. Experiments on both PASCAL-VOC 2012 and ADE20k benchmarks demonstrate the outstanding performance of our method. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Recommender System Combining Popularity and Novelty Based on One-Mode Projection of Weighted Bipartite Network

引用

Computers, Materials & Continua 2020年第4期63卷 489-507页

作者： Yong Yu Yongjun Luo Tong Li Shudong Li Xiaobo Wu Jinzhuo Liu Yu Jiang School of Software Key Laboratory in Software Engineering of Yunnan ProvinceYunnan UniversityKunming650091China School of Big Data Yunnan Agricultural UniversityKunming650201China Cyberspace Institute of Advanced Technology Guangzhou UniversityGuangzhou510006China School of Computer Science and Cyber Engineering Guangzhou UniversityGuangzhou510006China

Personalized recommendation algorithms,which are effective means to solve information overload,are popular topics in current *** this paper,a recommender system combining popularity and novelty(RSCPN)based on one-mode projection of weighted bipartite network is *** edge between a user and item is weighted with the item’s rating,and we consider the difference in the ratings of different users for an item to obtain a reasonable method of measuring the similarity between *** can be used in the same model for popularity and novelty recommendation by setting different parameter values and analyzing how a change in parameters affects the popularity and novelty of the recommender *** verify and compare the accuracy,diversity and novelty of the proposed model with those of other models,and results show that RSCPN is feasible.

关键词： Personalized recommendation one-mode projection weighted bipartite network novelty recommendation diversity

来源：评论

学校读者我要写书评

暂无评论

Non-Intrusive Load Monitoring for Multi-objects in Smart Building 4

Non-Intrusive Load Monitoring for Multi-objects in Smart Bui...

引用

4th International Balkan Conference on Communications and Networking, BalkanCom 2021

作者： Li, Dandan Li, Jiangfeng Zeng, Xin Stankovic, Vladimir Stankovic, Lina Shi, Qingjiang Tongji University School of Software Engineering Shanghai China University of Strathclyde Department of Electronic and Electrical Engineering GlasgowG1 1XW United Kingdom Tongji University College of Electronic and Information Engineering Shanghai China Shenzhen Research Institute of Big Data Shenzhen China

ISBN: (纸本)9781665402583

The rapidly expansion of Internet of Things (IoT) has ignited renewed interest in energy disaggregation via non-intrusive load monitoring (NILM). Compared to the more frequent NILM approach of training one model for each appliance, this paper proposes a multi-label learning approach based on the widely cited sequence2point convolutional neural network (CNN). Using the smart meter readings collected in an office building, we demonstrate the accuracy and practicality of the proposed network compared to start-of-the-art one-to-one NILM models. © 2021 IEEE.

关键词： Energy efficiency

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：