检索结果-内蒙古大学图书馆

SenAttack: adversarial attack method based on perturbation sensitivity and perceptual color distance

APPLIED INTELLIGENCE 2023年第23期53卷 28937-28953页

作者： Sun, Jiaze Long, Siyuan Ma, Xianyan Xian Univ Posts & Telecommun Sch Comp Sci & Technol Xian 710121 Peoples R China Network Data Anal & Intelligent Proc Shaanxi Key Lab Xian 710121 Peoples R China Xian Key Lab Big Data & Intelligent Comp Xian 710121 Peoples R China

Deep neural networks have effectively processed substantial quantities of image, text, and speech data. However, they are susceptible to adversarial sample attacks, where slight perturbations modify the input to generate malicious outputs. These attacks pose a severe security risk to deep neural network systems. Prior methods for generating adversarial samples rely on objective L-p norm to measure the distance between original and adversarial samples, which can be easily detected by humans. Moreover, these attacks treat perturbations of different pixels equally without considering the sensitivity of the human eye to different pixels. To address these challenges, this paper proposes SenAttack, an adversarial attack method based on perturbation sensitivity and perceptual color distance. SenAttack evaluates perceptual similarity using the perceptual color distance and combines the perturbation sensitivity of image structural information as an additional term in the calculation of the perceptual color distance. Furthermore, it employs an adaptive parameter adjustment algorithm to optimize the iterative process. Experimental results demonstrate that SenAttack achieves a high success rate in white-box attacks on Inception-v3 and ResNet-50 models with few iterations. The average norm distance and perceptual color distance of the adversarial samples are reduced by over 10% and 50% respectively, significantly enhancing their imperceptibility. SenAttack also exhibits superior visual effects in high confidence scenarios and improved robustness against defense models with partial image compression. SenAttack provides a novel approach to adversarial attacks by considering the sensitivity of the human eye and employing perceptual color distance. The results highlight its effectiveness in generating imperceptible adversarial samples and its potential for enhancing the security of deep neural network systems.

关键词： Adversarial sample Adaptive parameter adjustment Perturbation sensitivity Perceptual color distance

来源：评论

学校读者我要写书评

暂无评论

Pairwise open-sourced dataSet protection based on adaptive blind watermarking

引用

APPLIED INTELLIGENCE 2023年第14期53卷 17391-17410页

作者： Pang, Zilong Wang, Mingxu Cao, Lvchen Chai, Xiuli Gan, Zhihua Henan Univ Inst Intelligent Network Syst Intelligent Data Proc Engn Res Ctr Henan Prov Sch Software Kaifeng 475004 Peoples R China Henan Univ Sch Artificial Intelligence Henan Key Lab Big Data Anal & Proc Zhengzhou 450046 Peoples R China Henan Key Lab Cyberspace Situat Awareness Zhengzhou 450001 Peoples R China

The cost of collecting and labeling open-sourced datasets which promote the development of deep learning is expensive. Thus, it is important to design an efficient open-sourced data set protection algorithm to detect the open-sourced data sets used in illegal ways. In this paper, a protection algorithm based on adaptive robust blind-watermark is proposed suitable for multiple paired open-sourced datasets, and the evaluation criteria of the algorithm are defined. Specifically, in the embed stage, the highly concealed of the watermark is realized by combining UNet and double GAN to take into account the local and global features of the carrier and the watermark image. A preprocess network is used in the Embedding network to adapt different watermark size. In the extraction part, a modified feature sharing UNet with GAN is used to ensure robustness of the extraction network. Paired datasets are used for training to ensure accurate extraction of watermarks. After the target model is trained using the watermarked dataset, its inference output will contain watermark information. When it is believed that a suspicious model is illegally trained with the dataset, it can be verified by the watermark extracted from inference output of the suspicious model. We evaluate our method on three target models including nine datasets. The results show that our framework successfully verifies the dataset used illegally and without a noticeable impact on the target model task when training with the watermark dataset.

关键词： Blind watermark Deep learning Image processing Open-sourced dataset protection

来源：评论

学校读者我要写书评

暂无评论

Three-Channel Convolutional Neural network for Polarimetric SAR Images Classification

引用

IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING 2020年 13卷 4895-4907页

作者： Hua, Wenqiang Xie, Wen Jin, Xiaomin Xian Univ Posts & Telecommun Sch Comp Sci & Technol Shaanxi Key Lab Network Data Anal & Intelligent P Xian 710121 Peoples R China

Terrain classification is an important topic in polarimetric synthetic aperture radar (PolSAR) image processing and interpretation. A novel PolSAR classification method based on the three-channel convolutional neural network (Tc-CNN) is proposed and this method can effectively take the advantage of unlabeled samples to improve the performance of classification with a small number of labeled samples. Several strategies are included in the proposed method. First, in order to take the advantage of unlabeled samples, a data enhancement method based on the neighborhood nearest neighbor propagation method is proposed to enlarge the number of labeled samples. Second, to increase the role of central pixel in convolutional neural network classification based on the pixel, a spatial weighted method is proposed to increase the weight of central pixel features and weak the weight of other types of pixel features. Third, a specific deep model for PolSAR image classification (named Tc-CNN) is proposed, which can obtain more scale and deep polarization information to improve the classification results. The experimental results show that the proposed method achieves a much better performance than the existing classification methods when the number of labeled samples is few.

关键词： Convolutional neural network (CNN) polarimetric synthetic aperture radar (PolSAR) terrain classification three-channel convolutional neural network (Tc-CNN)

来源：评论

学校读者我要写书评

暂无评论

Consistency-oriented clustering ensemble via data reconstruction

引用

APPLIED INTELLIGENCE 2024年第20期54卷 9641-9654页

作者： Zhang, Hengshan Wang, Yun Chen, Yanping Sun, Jiaze Xian Univ Posts & Telecommun Sch Comp Sci & Technol Xian Peoples R China Shanxi Key Lab Network Data Anal & Intelligent Pro Xian Peoples R China Xian Key Lab Big Data & Intelligent Comp Xian Peoples R China

The study highlights that using different distance measures on the same dataset leads to varying clustering results, making the choice of distance measure a challenge when prior knowledge is lacking. To address this issue, a consistency-oriented clustering ensemble via data reconstruction is developed. This approach eliminates the need to select a specific distance measure and achieves higher consistency between the clustering ensemble and base clusterings while maintaining superior clustering performance. First, the base clustering is generated via the clustering with different distance measures and a consistency definition is introduced in the proposed method. Then the ensemble process updates the weights of base clusterings to ensure they reach the consistency. At the same time data reconstruction process is integrated into the ensemble process to guarantee a high convergence rate and efficient clustering. Finally, the clustering ensemble result is achieved with the higher consistency measure and improved clustering performance by balancing both factors. In the experiment, the effectiveness of the proposed method is verified and the specification of the parameters is advised through the various experimental outcomes.

关键词： Clustering ensemble Consistency data reconstruction Similarity matrix Distance measure

来源：评论

学校读者我要写书评

暂无评论

Polarimetric SAR Image Classification Based on Ensemble Dual-Branch CNN and Superpixel Algorithm

引用

IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING 2022年 15卷 2759-2772页

作者： Hua, Wenqiang Zhang, Cong Xie, Wen Jin, Xiaomin Xian Univ Posts & Telecommun Sch Comp Sci & Technol Shaanxi Key Lab Network Data Anal & Intelligent P Xian 710121 Peoples R China

Recently, convolutional neural networks (CNNs) have been successfully utilized in polarimetric synthetic aperture radar (PolSAR) image classification and obtained promising results. However, most CNN-based classification methods require a large number of labeled samples and it is difficult to obtain sufficient labeled samples. For this reason, an ensemble dual-branch CNN (EDb-CNN) is proposed for PolSAR image classification with small samples. First, to solve the problem of the small sample in PolSAR image classification, a new data enhancement method based on the superpixel algorithm is proposed to expand the number of labeled samples. Second, to obtain different scales of features from PolSAR images, a Db-CNN model is proposed. This model contains two parallel CNN structures. One CNN branch is used to extract the polarization features from the complex coherency matrix. The other branch is utilized to extract the spatial features based on weighted spatial neighborhood. On the top of these two branches, a feature fusion model is adopted to combine these two deep features, and a weighted loss function is employed to improve the learning procedure. Then, the ensemble learning algorithm is used for each CNN branch and Db-CNN network to obtain the better classification results. Finally, a postprocess algorithm based on the superpixel algorithm is proposed to improve the consistency of classification results. Experiments on two PolSAR datasets show that the proposed method achieves a much better performance than other classification methods, especially when only a few labeled samples are available.

关键词： Feature extraction Image classification Licenses Classification algorithms Covariance matrices Remote sensing Image recognition Dual-branch convolutional neural network (Db-CNN) polarimetric SAR superpixels terrain classification

来源：评论

学校读者我要写书评

暂无评论

RetinexGAN:Unsupervised Low-Light Enhancement With Two-Layer Convolutional Decomposition networks

引用

IEEE ACCESS 2021年 9卷 56539-56550页

作者： Ma, Tian Guo, Ming Yu, Zhenhua Chen, Yanping Ren, Xincheng Xi, Runtao Li, Yuancheng Zhou, Xinlei Xian Univ Sci & Technol Coll Comp Sci & Technol Xian 710054 Peoples R China Xian Univ Posts & Telecommun Shaanxi Key Lab Network Data Anal & Intelligent P Xian 710121 Peoples R China Yanan Univ Shaanxi Key Lab Intelligent Proc Big Energy Data Yanan 716000 Peoples R China

In the field of image enhancement, using deep learning methods to enhance low-light images is currently mainstream. However, the methods often have complex network structures of a large number of parameters, and their training often uses paired data-sets, which are difficult to obtain in actual practice. To solve these problems, this paper proposes a simple generative adversarial network and Retinex model, dubbed RetinexGAN, that is completely trained using unpaired data-sets. It contains a decomposition network and two discriminator networks. To reduce the parameters of the network, only two convolution layers are used in the decomposition network. We show more challenging testing data where some parts of the image are underexposed and others are normal light. Both quantitative and visual results show that RetinexGAN is largely superior to state-of-the-art methods.

关键词： Lighting Image enhancement Training Histograms Gallium nitride Convolution Generative adversarial networks Low-light image enhancement generative adversarial networks two-layer convolutional networks

来源：评论

学校读者我要写书评

暂无评论

CIE-LSCP: color image encryption scheme based on the lifting scheme and cross-component permutation

引用

COMPLEX & intelligent SYSTEMS 2023年第1期9卷 927-950页

作者： Chen, Xiuhui Gong, Mengxin Gan, Zhihua Lu, Yang Chai, Xiuli He, Xin Henan Univ Sch Artificial Intelligence Henan Key Lab Big Data Anal & Proc Kaifeng 475004 Peoples R China Henan Univ Inst Intelligent Network Syst Sch Software Intelligent Data Proc Engn Res Ctr Henan Prov Kaifeng 475004 Peoples R China

Encryption of meaningful images into unidentifiable noise-like images can improve the security of images during storage and transmission. In this paper, a novel color image encryption method based on the lifting scheme and cross-component permutation (CIE-LSCP) is introduced. First, original image is divided into red, green and blue components, and then the three components are processed by a preprocessing strategy based on the lifting scheme (PSLS) to change the statistical distribution of the pixels. Second, a block-based cross-component permutation based on index vectors (BCPIV) is presented to divide three preprocessed components into blocks and perform the cross-component permutation operation on them, and three random matrices are utilized to determine the target component, target block and target pixel position of the current pixel to be moved, respectively. Subsequently, a multi stochastic diffusion based on random sequences (MSDRS) is developed to alter the pixel values of the permutated components, and finally a ciphertext image is gotten by merging the three components. Moreover, the SHA256 hash values of the plaintext image are used to obtain the initial parameters of the chaotic system, and the obtained chaotic sequences are applied in the image encryption process. Wherein the generated random sequences are highly reliant on the plaintext image, making the encryption scheme resistant to both known-plaintext attacks and chosen-plaintext attacks. Experimental results demonstrate that the proposed scheme has good security and effectiveness and can be applied for secure transmission of digital images over the Internet.

关键词： Color image encryption Chaotic system The lifting scheme Cross-component permutation Multi stochastic diffusion

来源：评论

学校读者我要写书评

暂无评论

Cloud-decryption-assisted image compression and encryption based on compressed sensing

引用

MULTIMEDIA TOOLS AND APPLICATIONS 2022年第12期81卷 17401-17436页

作者： Fu, Jiangyu Gan, Zhihua Chai, Xiuli Lu, Yang Henan Univ Sch Artificial Intelligence Kaifeng 475004 Peoples R China Henan Univ Intelligent Data Proc Engn Res Ctr Henan Prov Sch Software Inst Intelligent Network Syst Kaifeng 475004 Peoples R China Henan Univ Henan Key Lab Big Data Anal & Proc Kaifeng 475004 Peoples R China

In this paper, we propose a new image compression and encryption scheme (ICES) based on compressed sensing (CS) and double random phase encoding (DRPE), as well as a joint decryption on the cloud and user side. In the encryption, the plain image is firstly decomposed into approximate component and detail components by discrete wavelet transform (DWT), then the approximate component is scrambled, and detail components are compressed using a measurement matrix constructed by the Logistic-tent system. Secondly, the scrambled approximate component and compressed detail components are combined into a complex matrix, and then it is subjected to DRPE to get the amplitude and phase matrices. Subsequently, the resulting matrices are performed random pixel scrambling and diffusion to obtain the final cipher image. During the decryption, the cipher image is first partially decrypted on the cloud, and then fully decrypted to recover the original image on the user side, and we can judge whether the cloud has cheated us by comparing the contour similarity between the approximate component and the detail component returned by the cloud to the user, which not only significantly shortens the decryption time, but also effectively prevents malicious deception of the cloud. Experimental results and performance analyses demonstrate its effectiveness and efficiency.

关键词： Compressed sensing Cloud Image compression and encryption schemes Double random phase encoding

来源：评论

学校读者我要写书评

暂无评论

Efficient capacity-distortion reversible data hiding based on combining multipeak embedding with local complexity

引用

APPLIED INTELLIGENCE 2022年第11期52卷 13006-13026页

作者： Fu, Zhifeng Gong, Mengxin Long, Guoqiang Gan, Zhihua Chai, Xiuli Lu, Yang Henan Univ Sch Artificial Intelligence Zhengzhou 450046 Peoples R China Henan Univ Henan Key Lab Big Data Anal & Proc Kaifeng 475004 Peoples R China Henan Univ Inst Intelligent Network Syst Sch Software Intelligent Data Proc Engn Res Ctr Henan Prov Kaifeng 475004 Peoples R China

Recently, most histogram shifting-based reversible data hiding (RDH) algorithms have considered the impact of textural information on embedding performance, while exploiting pixels with different local complexities. Prioritizing pixels with small local complexities to accommodate secret data decreases the invalid shifting pixels, thereby reducing distortion. However, though effective, the local complexity is not calculated precisely enough, which results in inaccurate texture division and does not considerably reduce distortion. Thus, we employ a novel local complexity calculation and multipeak embedding (MPE) to effectively improve the capacity-distortion performance. Specifically, the host image is first preprocessed by the dot-cross pattern and divided into two subsets. Then the pixel local complexity of each subset is computed by using the spatial correlation of pixels (SCOP) to improve calculation accuracy. Finally, the peak bins to be expanded in the regions with lower local complexity are adaptively selected for embedding with secret data by MPE. To ensure that the authorized operator can securely gain error-free secret data, we design the location index sequences as special keys, which guarantee the algorithm reversibility and enhance the security of the algorithm at the same time. Experimental results show that our algorithm has superior embedding effectiveness compared to some state-of-the-art RDH methods.

关键词： Reversible data hiding Invalid shifting pixels Local complexity Location index mechanism Multipeak embedding

来源：评论

学校读者我要写书评

暂无评论

A robust compressed sensing image encryption algorithm based on GAN and CNN

引用

JOURNAL OF MODERN OPTICS 2022年第2期69卷 103-120页

作者： Chai, Xiuli Tian, Ye Gan, Zhihua Lu, Yang Wu, Xiang-Jun Long, Guoqiang Henan Univ Sch Artificial Intelligence Kaifeng 475004 Peoples R China Henan Univ Henan Key Lab Big Data Anal & Proc Kaifeng 475004 Peoples R China Henan Univ Intelligent Data Proc Engn Res Ctr Henan Prov Sch Software Inst Intelligent Network Syst Kaifeng Peoples R China

In this paper, a robust compressed sensing image encryption algorithm based on generative adversarial network, convolutional neural network (CNN) denoising network and chaotic system is developed. Firstly, we use a sampling network to get the measurement of plain image. Second, the cipher image is obtained by scrambling the measurement through the Logistic-Tent chaotic system. After getting cipher image, the decryption party obtains decrypted measurement by inverse scrambling the cipher image, and then sends it to the reconstruction network to obtain decrypted reconstructed image. Finally, by using the CNN denoiser, the image quality and the visual expression of final decrypted image can be improved. In this scheme, the dual denoiser structure based on reconstruction network and CNN denoiser can effectively resist noise attacks. Besides, the proposed training strategy with noise injection can further improve the robustness of network. Experiments show our method has high reconstruction quality, efficiency, robustness and security.

关键词： Compressed sensing image encryption generative adversarial network (GAN) convolutional neural network (CNN) deep learning robustness

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：