检索结果-内蒙古大学图书馆

ASAG: Building Strong One-Decoder-Layer Sparse Detectors via Adaptive Sparse Anchor Generation

学校读者我要写书评

暂无评论

ASAG: Building Strong One-Decoder-Layer Sparse Detectors via...

International Conference on computer Vision (ICCV)

作者： Shenghao Fu Junkai Yan Yipeng Gao Xiaohua Xie Wei-Shi Zheng School of Computer Science and Engineering Sun Yat-sen University China Guangdong Province Key Laboratory of Information Security Technology China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China Pengcheng Lab China

Recent sparse detectors with multiple, e.g. six, decoder layers achieve promising performance but much inference time due to complex heads. Previous works have explored using dense priors as initialization and built one-decoder-layer detectors. Although they gain remarkable acceleration, their performance still lags behind their six-decoder-layer counterparts by a large margin. In this work, we aim to bridge this performance gap while retaining fast speed. We find that the architecture discrepancy between dense and sparse detectors leads to feature conflict, hampering the performance of one-decoder-layer detectors. Thus we propose Adaptive Sparse Anchor Generator (ASAG) which predicts dynamic anchors on patches rather than grids in a sparse way so that it alleviates the feature conflict problem. For each image, ASAG dynamically selects which feature maps and which locations to predict, forming a fully adaptive way to generate image-specific anchors. Further, a simple and effective Query Weighting method eases the training instability from adaptiveness. Extensive experiments show that our method outperforms dense-initialized ones and achieves a better speed-accuracy trade-off. The code is availab.e at https://***/iSEE-lab.ratory/ASAG.

关键词：

Estimator Meets Equilibrium Perspective: A Rectified Straight Through Estimator for Binary Neural Networks Training

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wu, Xiao-Ming Zheng, Dian Liu, Zuhao Zheng, Wei-Shi School of Computer Science and Engineering Sun Yat-sen University China Pengcheng Lab China Guangdong Province Key Laboratory of Information Security Technology China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China

关键词： Stability

Deep Feature Surgery: Towards Accurate and Efficient Multi-Exit Networks

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Gong, Cheng Chen, Yao Luo, Qiuyang Lu, Ye Li, Tao Zhang, Yuzhi Sun, Yufei Zhang, Le College of Software Nankai University China College of Computer Science Nankai University China National University of Singapore Singapore School of Information and Communication Engineering University of Electronic Science and Technology of China China HAIHE Lab of ITAI China Tianjin Key Laboratory of Network and Data Security Technology China Key Laboratory of Data and Intelligent System Security Ministry of Education

Multi-exit network is a promising architecture for efficient model inference by sharing backbone networks and weights among multiple exits. However, the gradient conflict of the shared weights results in sub-optimal accuracy. This paper introduces Deep Feature Surgery (DFS), which consists of feature partitioning and feature referencing approaches to resolve gradient conflict issues during the training of multi-exit networks. The feature partitioning separates shared features along the depth axis among all exits to alleviate gradient conflict while simultaneously promoting joint optimization for each exit. Subsequently, feature referencing enhances multi-scale features for distinct exits across varying depths to improve the model accuracy. Furthermore, DFS reduces the training operations with the reduced complexity of backpropagation. Experimental results on Cifar100 and ImageNet datasets exhibit that DFS provides up to a 50.00% reduction in training time and attains up to a 6.94% enhancement in accuracy when contrasted with baseline methods across diverse models and tasks. Budgeted batch classification evaluation on MSDNet demonstrates that DFS uses about 2× fewer average FLOPs per image to achieve the same classification accuracy as baseline methods on Cifar100. The code is availab.e at https: //***/GongCheng1919/dfs. Copyright © 2024, The Authors. All rights reserved.

关键词： Budget control

Evaluation of Age of information for LDPC Coded Transmission over AWGN Channels

学校读者我要写书评

暂无评论

Evaluation of Age of Information for LDPC Coded Transmission...

IEEE Vehicular technology Conference

作者： Mangang Xie Qianfan Wang Jie Gong Xiao Ma School of Data and Computer Science Sun Yat-sen University Guangzhou China School of Electronics and Communication Engineering Sun Yat-sen University Guangzhou China Guangdong Key Laboratory of Information Security Technology Sun Yat-sen University Guangzhou China

Age of information (AoI) is an important metric in real-time status update communication system to assess the freshness of information. Different from previous works, this paper focuses on the average AoI over additive white Gaussian noise (AWGN) channels. A fixed redundancy (FR) coding scheme is considered, which encodes each k-bits update as an n-bits packet by a low-density parity-check (LDPC) code. By using the renewal-reward theory, a closed-form expression of the average AoI under the FR scheme is derived. Simulation results show that the average AoI relies on E_b/N_0, especially in the low E_b/N_0 region. For different E_b/N_0, an optimal code length always exists to minimize the average AoI. In addition, the same average AoI can be achieved with different code lengths in the high E_b/N_0 region. Hence, in the high E_b/N_0 region, short codes are preferred especially when the transmission delay is taken into account.

关键词： Encoding Parity check codes AWGN channels Monitoring Real-time systems Binary phase shift keying Measurement

Person re-identification by contour sketch under moderate clothing change

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Yang, Qize Wu, Ancong Zheng, Wei-Shi School of Data and Computer Science Sun Yat-sen University Guangzhou510275 China School of Electronics and Information Technology Sun Yat-sen University Guangzhou510275 China Guangdong Province Key Laboratory of Information Security China School of Data and Computer Science Sun Yatsen University Guangzhou510275 China Peng Cheng Laboratory Shenzhen518005 China Ministry of Education China

Person re-identification (re-id), the process of matching pedestrian images across different camera views, is an important task in visual surveillance. Substantial development of re-id has recently been observed, and the majority of existing models are largely dependent on color appearance and assume that pedestrians do not change their clothes across camera views. This limitation, however, can be an issue for re-id when tracking a person at different places and at different time if that person (e.g., a criminal suspect) changes his/her clothes, causing most existing methods to fail, since they are heavily relying on color appearance and thus they are inclined to match a person to another person wearing similar clothes. In this work, we call the person re-id under clothing change the "cross-clothes person re-id". In particular, we consider the case when a person only changes his clothes moderately as a first attempt at solving this problem based on visible light images;that is we assume that a person wears clothes of a similar thickness, and thus the shape of a person would not change significantly when the weather does not change substantially within a short period of time. We perform cross-clothes person re-id based on a contour sketch of person image to take advantage of the shape of the human body instead of color information for extracting features that are robust to moderate clothing change. To select/sample more reliable and discriminative curve patterns on a body contour sketch, we introduce a learning-based spatial polar transformation (SPT) layer in the deep neural network to transform contour sketch images for extracting reliable and discriminant convolutional neural network (CNN) features in a polar coordinate space. An angle-specific extractor (ASE) is applied in the following layers to extract more fine-grained discriminant angle-specific features. By varying the sampling range of the SPT, we develop a multistream network for aggregating multi-granular

关键词： Color

A Four-Pronged Defense Against Byzantine Attacks in Federated Learning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wan, Wei Hu, Shengshan Li, Minghui Lu, Jianrong Zhang, Longling Zhang, Leo Yu 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

Federated learning (FL) is a nascent distributed learning paradigm to train a shared global model without violating users' privacy. FL has been shown to be vulnerable to various Byzantine attacks, where malicious participants could independently or collusively upload well-crafted updates to deteriorate the performance of the global model. However, existing defenses could only mitigate part of Byzantine attacks, without providing an all-sided shield for FL. It is difficult to simply combine them as they rely on totally contradictory assumptions. In this paper, we propose FPD, a four-pronged defense against both non-colluding and colluding Byzantine attacks. Our main idea is to utilize absolute similarity to filter updates rather than relative similarity used in existingI works. To this end, we first propose a reliable client selection strategy to prevent the majority of threats in the bud. Then we design a simple but effective score-based detection method to mitigate colluding attacks. Third, we construct an enhanced spectral-based outlier detector to accurately discard abnormal updates when the training data is not independent and identically distributed (non-IID). Finally, we design update denoising to rectify the direction of the slightly noisy but harmful updates. The four sequentially combined modules can effectively reconcile the contradiction in addressing non-colluding and colluding Byzantine attacks. Extensive experiments over three benchmark image classification datasets against four state-of-the-art Byzantine attacks demonstrate that FPD drastically outperforms existing defenses in IID and non-IID scenarios (with 30% improvement on model accuracy). © 2023, CC BY.

关键词： Classification (of information)

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

学校读者我要写书评

暂无评论

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

JPEG steganalysis based on DenseNet

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Yang, Jianhua Shi, Yun-Qing Wong, Edward K. Kang, Xiangui Guangdong Key Lab of Information Security School of Data and Computer Science Sun Yat-Sen University Guangzhou510006 China Department of ECE New Jersey Institute of Technology NewarkNJ07102 United States Department of Computer Science and Engineering New York University Tandon School of Engineering BrooklynNY11201 United States

Different from the conventional deep learning work based on an images content in computer vision, deep steganalysis is an art to detect the secret information embedded in an image via deep learning, pose challenge of detection weak information invisible hidden in a host image thus learning in a very low signal-to-noise (SNR) case. In this paper, we propose a 32- layer convolutional neural Networks (CNNs) in to improve the efficiency of preprocess and reuse the features by concatenating all features from the previous layers with the same featuremap size, thus improve the flow of information and gradient. The shared features and bottleneck layers further improve the feature propagation and reduce the CNN model parameters dramatically. Experimental results on the BOSSbase, BOWS2 and ImageNet datasets have showed that the proposed CNN architecture can improve the performance and enhance the robustness. To further boost the detection accuracy, an ensemble architecture called as CNN-SCA-GFR is proposed, CNN-SCAGFR is also the first work to combine the CNN architecture and conventional method in the JPEG domain. Experiments show that it can further lower detection errors. Compared with the state-of-the-art method XuNet [1] on BOSSbase, the proposed CNN-SCA-GFR architecture can reduce detection error rate by 5.67% for 0.1 bpnzAC and by 4.41% for 0.4 bpnzAC while the number of training parameters in CNN is only 17% of what used by XuNet. It also decreases the detection errors from the conventional method SCA-GFR by 7.89% for 0.1 bpnzAC and 8.06% for 0.4 bpnzAC, respectively. Copyright © 2017, The Authors. All rights reserved.

关键词： Steganography

Spatial image steganography based on generative adversarial network

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Yang, Jianhua Liu, Kai Kang, Xiangui Wong, Edward K. Shi, Yun-Qing Guangdong Key Lab of Information Security School of Data and Computer Science Sun Yat-Sen University Guangzhou510006 China Department of Computer Science and Engineering New York University Tandon School of Engineering BrooklynNY11201 United States Department of ECE New Jersey Institute of Technology NewarkNJ07102 United States

With the recent development of deep learning on steganalysis, embedding secret information into digital images faces great challenges. In this paper, a secure steganography algorithm by using adversarial training is proposed. The architecture contain three component modules: a generator, an embedding simulator and a discriminator. A generator based on U-NET to translate a cover image into an embedding change probability is proposed. To fit the optimal embedding simulator and propagate the gradient, a function called Tanh-simulator is proposed. As for the discriminator, the selection-channel awareness (SCA) is incorporated to resist the SCA based steganalytic methods. Experimental results have shown that the proposed framework can increase the security performance dramatically over the recently reported method ASDL-GAN, while the training time is only 30% of that used by ASDL-GAN. Furthermore, it also performs better than the hand-crafted steganographic algorithm S-UNIWARD. Copyright © 2018, The Authors. All rights reserved.

关键词： Generative adversarial networks