检索结果-内蒙古大学图书馆

DIG: Improved DINO for Graffiti Detection

IEEE Transactions on Systems, Man, and Cybernetics: Systems 2025年第5期55卷 3557-3569页

作者： Wang, Bingshu Mao, Qianchen Liu, Aifei Chen, Long Chen, C. L. Philip Northwestern Polytechnical University School of Software Xi’an710129 China Shenzhen University Guangdong Key Laboratory of Intelligent Information Processing Shenzhen Key Laboratory of Media Security Shenzhen518060 China University of Macao Department of Computer and Information Science China South China University of Technology School of Computer Science and Engineering Guangzhou510641 China

Graffiti detection is essential in historic building protection and urban neighborhood management. Graffiti detection has made significant progress in recent years based on the development of deep learning. However, small-scale graffiti, interference from the background, and the false detection of word parts in graffiti make it a challenging problem. This article proposes a Transformer-based high-precision graffiti detection method, namely DIG. Precisely, it consists of three modules: 1) Spatial query selection (SQS), scale-aware IoU loss (SIL);2) Denoising Task with binary contrastive denoising (BCDN);and 3) IoU-guided box denoising (IBD) modules. To detect small-scale graffiti, this paper proposes SIL to help the loss function to perceive small-scale graffiti and large-scale graffiti fairly. To reduce the false detection of word parts, this article presents the SQS module, which integrates spatial information into the query selection process of the Encoder to filter out falsely detected bounding boxes within the graffiti. To reduce the interference from the background, this article introduces a denoising task with BCDN and IBD modules, improving the model’s ability to distinguish graffiti from the background and accurately select appropriate bounding boxes. A large number of experimental results on the STORM dataset show that our method achieves state-of-the-art results with an AP50 of 87.9%. Moreover, DIG achieved competitive results on the FineFM dataset for mask detection. This indicates that DIG can also be conveniently transferred to detect other scenarios. © 2013 IEEE.

关键词： Historic preservation

来源：评论

学校读者我要写书评

暂无评论

Lip-Audio Modality Fusion for Deep Forgery Video Detection

引用

computers, Materials & Continua 2025年第2期82卷 3499-3515页

作者： Yong Liu Zhiyu Wang Shouling Ji Daofu Gong Lanxin Cheng Ruosi Cheng College of Cyberspace Security Information Engineering UniversityZhengzhou450001China Research Institute of Intelligent Networks Zhejiang LabHangzhou311121China College of Computer Science and Technology Zhejiang UniversityHangzhou310027China Henan Key Laboratory of Cyberspace Situation Awareness Zhengzhou450001China Key Laboratory of Cyberspace Security Ministry of EducationZhengzhou450001China

In response to the problem of traditional methods ignoring audio modality tampering, this study aims to explore an effective deep forgery video detection technique that improves detection precision and reliability by fusing lip images and audio signals. The main method used is lip-audio matching detection technology based on the Siamese neural network, combined with MFCC (Mel Frequency Cepstrum Coefficient) feature extraction of band-pass filters, an improved dual-branch Siamese network structure, and a two-stream network structure design. Firstly, the video stream is preprocessed to extract lip images, and the audio stream is preprocessed to extract MFCC features. Then, these features are processed separately through the two branches of the Siamese network. Finally, the model is trained and optimized through fully connected layers and loss functions. The experimental results show that the testing accuracy of the model in this study on the LRW (Lip Reading in the Wild) dataset reaches 92.3%;the recall rate is 94.3%;the F1 score is 93.3%, significantly better than the results of CNN (Convolutional Neural networks) and LSTM (Long Short-Term Memory) models. In the validation of multi-resolution image streams, the highest accuracy of dual-resolution image streams reaches 94%. Band-pass filters can effectively improve the signal-to-noise ratio of deep forgery video detection when processing different types of audio signals. The real-time processing performance of the model is also excellent, and it achieves an average score of up to 5 in user research. These data demonstrate that the method proposed in this study can effectively fuse visual and audio information in deep forgery video detection, accurately identify inconsistencies between video and audio, and thus verify the effectiveness of lip-audio modality fusion technology in improving detection performance.

关键词： Deep forgery video detection lip-audio modality fusion mel frequency cepstrum coefficient siamese neural network band-pass filter

来源：评论

学校读者我要写书评

暂无评论

Malware Detection Method Based on TPE Optimized Stacking 3

Malware Detection Method Based on TPE Optimized Stacking

引用

3rd International Conference on High Performance Computing and Communication engineering, HPCCE 2023

作者： Zhao, Dongbo Yuan, Zesong Xie, Nannan Jilin Province Key Laboratory of Network and Information Security School of Computer Science and Technology Changchun University of Science and Technology Changchun130022 China

ISBN: (数字)9781510674653

ISBN: (纸本)9781510674646

Fast detection and categorization of malware are increasingly important for securing hosts and networks. Although many Machine Learning models have been utilized to detect malware, single-model detection may not remain efficient in the face of diverse datasets. The continuously increasing size of malware need the distributed storage systems and distributed computing. The distributed methods and ensemble learning are introduced into malware detection, and an optimal combination of base classifiers suitable for stacking are designed. A hyperparameter optimization method based on the Tree-structured Parzen Estimator (TPE) approach to enhance malware detection. The proposed method is implemented on Apache Spark and Hadoop Distributed File System(HDFS). Experiments conducted on four independent datasets, including Android and Windows, demonstrate that the proposed method can achieve 99.41% accuracy on the Android dataset and 96.96% accuracy on the Windows dataset. © 2024 SPIE.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

A Location Recommendation Model Based on User Behavior and Sequence Influence 1

引用

8th International Conference on Internet on Things, ICIOT 2023, held as part of Services Conference Federation SCF 2023

作者： Mao, Weixuan Wen, Ruwang Cheng, Zhengxiang Mei, Zhuolin Wang, Haibin Wu, Bin Shi, Jiaoli School of Computer and Big Data Science Jiangxi Jiujiang China Jiujiang Key Laboratory of Network and Information Security Jiangxi Jiujiang China Institute of Information Security Jiujiang University Jiangxi Jiujiang China

ISBN: (数字)9783031517341

ISBN: (纸本)9783031517334

There are evident sequential patterns in human visits to different locations in daily life, which has led to the emergence of numerous sequence-based geographic location recommendation methods. However, previous research has mainly focused on considering the sequential patterns of check-ins as a whole, while overlooking the fact that users’ interactive behaviors at check-in locations are likely to influence their preferences for those locations, and subsequently impact their future check-in decisions. The problem of how to fully leverage a user’s overall location sequence and location interaction data to provide more personalized recommendations is still unresolved. To address this issue, we propose a geographic location recommendation model called UBSI (User Behavior and Sequence Influence) in this article. Our model is constructed based on check-in sequence, weight model, collaborative filtering, and n th-order additive Markov chain. First, UBSI collects users’ check-in sequence and interaction behavior data, generates interest weight using a weight model, integrates personalized data using collaborative filtering method, and finally applies the weighted n th-order additive Markov chain to mine sequential pattern of user check-ins for location recommendations. Finally, we conducted extensive experiments to demonstrate the effectiveness of our approach. The experimental results indicate that our UBSI solution can effectively improve the accuracy of recommendations. © 2024, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Collaborative filtering

来源：评论

学校读者我要写书评

暂无评论

EVPIR: Efficient and Verifiable Privacy-Preserving Image Retrieval in Cloud-assisted Internet of Things

引用

IEEE Internet of Things Journal 2025年第13期12卷 24259-24274页

作者： Li, Mingyue Li, Yuntao Du, Ruizhong Jia, Chunfu Shao, Wei Hebei University School of Cyber Security and Computer Baoding071000 China Key Lab on High Trusted Information Sys tem in Hebei Province Hebei China College of Cyber Sciences Nankai university Tianjin300350 China Tianjin Key Laboratory of Network and Data Security Technology Tianjin300350 China Jinan250014 China

With the proliferation of mobile devices and the advancement of cloud computing capabilities, cloud-assisted Internet of Things (IoT) attracts increased attention based on its computational and storage advantages. Upon these conveniences, there also raises privacy concerns that numerous solutions have been proposed to solve it. However, existing methods suffer from challenges such as low retrieval accuracy, inefficiency in large-scale image retrieval, and lack of efficient result verification. In this paper, we propose an Efficient Verifiable Privacy-preserving Image Retrieval scheme (EVPIR). Specifically, we design a hierarchical graph index to significantly enhance retrieval efficiency, which organizes image feature vectors into a multi-level structure, establishing connections between neighboring nodes within each layer and creating a highly structured and efficient retrieval framework. During the retrieval process, we employ greedy search algorithm to navigate these connections and identify the closest neighbors across different levels, which make the proposed multi-level approach reduce the search space at each level, achieving faster and more accurate retrieval. Furthermore, we design an efficient dynamic verifiable framework leveraging Chameleon hash functions and BLS signatures where we utilize Chameleon hash nodes based on Merkle Hash Trees (MHT) to enable dynamic updates of the verification tree and employ BLS signatures to construct multiple verification nodes for effectively shortening the verification path. Finally, security analysis shows that EVPIR can defend various threat models and extensive experiments further demonstrates that EVPIR can improve retrieval and verification efficiency. (Figure presented) © 2014 IEEE.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

Reinforcement learning of non-additive joint steganographic embedding costs with attention mechanism

引用

Science China(information Sciences) 2023年第3期66卷 273-286页

作者： Weixuan TANG Bin LI Weixiang LI Yuangen WANG Jiwu HUANG Institute of Artificial Intelligence and Blockchain Guangzhou University Guangdong Key Laboratory of Intelligent Information Processing Shenzhen Key Laboratory of Media Security Shenzhen University Shenzhen Institute of Artificial Intelligence and Robotics for Society School of Computer Science and Cyber Engineering Guangzhou University

Image steganography is the art and science of secure communication by concealing information within digital images. In recent years, the techniques of steganographic cost learning have developed rapidly. Although the existing methods can learn satisfactory additive costs, the interplay of different pixels' embedding impacts has not been considered, so the potential of learning may not be fully exploited. To overcome this limitation, in this paper, a reinforcement learning paradigm called Jo Po L(joint policy learning) is proposed to extend the idea of additive cost learning to a non-additive situation. Jo Po L aims to capture the interactions within pixel blocks by defining embedding policies and evaluating contributions of embedding impacts on a block level rather than a pixel level. Then, a policy network is utilized to learn optimal joint embedding policies for pixel blocks through interactions with the environment. Afterwards,these policies can be converted into joint embedding costs for practical message embedding. The structure of the policy network is designed with an effective attention mechanism and incorporated with the domain knowledge derived from traditional non-additive steganographic methods. The environment is responsible for assigning rewards according to the impacts of the sampled joint embedding actions, which are evaluated by the gradient information of a neural network-based steganalyzer. Experimental results show that the proposed non-additive method Jo Po L significantly outperforms the existing additive methods against both feature-based and CNN-based steganalzyers over different payloads.

关键词： information hiding non-additive steganography steganalysis cost learning image processing

来源：评论

学校读者我要写书评

暂无评论

A New High-Capacity Robust Image Steganography Method via Adversarial network 5

A New High-Capacity Robust Image Steganography Method via Ad...

引用

5th International Academic Exchange Conference on Science and Technology Innovation, IAECST 2023

作者： Dong, Weina Jiang, Zongbao Chen, Lifeng Sun, Wenquan University of PAP Key Laboratory of Network and Information Security of PAP Engineering Shaanxi Xi'an710086 China

ISBN: (纸本)9798350357738

Aiming at the problem that the high-capacity steganography model based on encoding-decoding network had weak robustness and could not resist noise attack and channel compression, an improved high-capacity robust image steganography algorithm based on adversarial network was proposed. In the proposed algorithm, encoder, decoder and critic based on densely connected convolutional networks (DenseNet) were designed. The secret information and the carrier image are jointly encoded into a steganographic image by the encoder, then the secret information was extracted by the decoder. The critic was used to discriminate between carrier images and steganographic images. A noise layer was added between encoder and decoder, Dropout, JPEG compression, gaussian blur, gaussian noise and salt and pepper noise were used to simulate a real environment of various kinds of noise attack. The steganography output by encoder was processed with different kinds of noise and decoded by decoder. Through model was trained, the message could be extracted from the noise-processed steganographic image by the decoder, so the ability to resist noise attacks was realized. The experiment results show that the steganographic capacity of the proposed algorithm reaches 0.450.95 bits per pixel (bpp) on 360 × 360 pixel images, which are 3.6 times higher than that of the earlier robust steganographic algorithms. The decoding accuracy reaches 0.72-0.97, which are 42% points higher than that of steganography without noise layer. The proposed algorithm not only guarantees high embedding quantity and high coding image quality, but also has stronger anti-noise capability. © 2023 IEEE.

关键词： Steganography

来源：评论

学校读者我要写书评

暂无评论

Reversible Data Hiding in Encrypted Images based on Classic McEliece Cryptosystem 2

Reversible Data Hiding in Encrypted Images based on Classic ...

引用

2nd International Conference on Image Processing, computer Vision and Machine Learning, ICICML 2023

作者： Jiang, Zongbao Zhang, Minqing Di, Fuqiang Dong, Weina University of PAP Key Laboratory of Network and Information Security of PAP Engineering Shaanxi Xi'an710086 China

ISBN: (纸本)9798350331417

Aiming at the problems that the security of traditional data hiding algorithms is difficult to be guaranteed and the existing algorithms' embedded volume is restricted under quantum operation attacks, a reversible data hiding algorithm in encrypted images based on Classic McEliece cryptosystem is proposed. The image pixels are first decomposed into bit planes to generate a data stream, and then encrypted in groups using the Classic McEliece cipher algorithm. In the encryption process, the fault tolerance mechanism in its encryption algorithm is utilized to select the appropriate modulation noise composition and the mapping relationship between the secret data, and the error pattern is modulated into noise to interfere with the encryption process, so as to realize the embedding of the secret data. Finally, the marked encrypted image is obtained by using the exclusive-or encryption and scrambling encryption processes. The experimental results show that the algorithm has the reversibility of carrier recovery. Compared with the existing algorithms, this algorithm's embedded volume is advantageous, and its highest embedding rate is 4.43548 bpp, which realizes large-capacity data hiding while improving security. © 2023 IEEE.

关键词： Classic McEliece error-correcting codes high-capacity hiding reversible data hiding in encrypted images

来源：评论

学校读者我要写书评

暂无评论

A two Stage Mixed Integer Programming Model for Distributionally Robust State Based Non-Intrusive Load Monitoring

引用

IEEE Transactions on Consumer Electronics 2025年第1期71卷 1024-1033页

作者： Zhang, Chen Chai, Zimo Yang, Linfeng University of Shanghai for Science and Technology School of Mechanical Engineering Shanghai China Guangxi University School of Computer Electronics and Information the Guangxi Key Laboratory of Multimedia Communication and Network Technology Nanning China

This paper presents a non-intrusive load monitoring (NILM) model based on two-stage mixed-integer linear programming theory. Compared with other mixed integer optimization-based models, this paper model introduces fewer integer variables and richer absolute error function of load decomposition, which makes the power state selection of each device more accurate and the power consumption more accurate fitting. First, to tackle the issues related to noisy load data, an innovative load feature extraction model based on Kullback-Leibler distributionally robust optimization principles is introduced. Then the key features (power boundary/fluctuation features) of each device identified through this robust model are integrated into the constraints of the two-stage NILM model. The two-stage complementary framework includes: the determination of device state interval in the first stage;and the accurate fitting of device power consumption within the device state interval in the second stage. Comparative validation against existing optimization-based models on the AMPds, REFIT, and actual laboratory data sets demonstrate that our proposed model significantly enhances power decomposition accuracy and computational efficiency. In addition, the two-stage complementary framework and load feature extraction model can be applied to other optimization-based models of NILM to improve the computational efficiency of each model and the accuracy of load decomposition. © 2025 IEEE.

关键词： Mixed-integer linear programming

来源：评论

学校读者我要写书评

暂无评论

Shorter ZK-SNARKs from square span programs over ideal lattices

引用

Cybersecurity 2025年第1期8卷 208-226页

作者： Xi Lin Heyang Cao Feng‑Hao Liu Zhedong Wang Mingsheng Wang Key Laboratory of Cyberspace Security Defense Institute of Information EngineeringChinese Academy of SciencesBeijingChina School of Cyber Security University of Chinese Academy of SciencesBeijingChina School of Electrical Engineering&Computer Science Washington State UniversityPullmanWAUSA School of Cyber Science and Engineering Shanghai Jiaotong UniversityShanghaiChina

Zero-knowledge succinct non-interactive arguments of knowledge(zk-SNARKs)are cryptographic protocols that ofer efcient and privacy-preserving means of verifying NP language relations and have drawn considerable atten‑tion for their appealing applications,e.g.,verifable computation and anonymous payment *** with the pre-quantum case,the practicability of this primitive in the post-quantum setting is still unsatisfactory,espe‑cially for the space *** tackle this issue,this work seeks to enhance the efciency and compactness of lat‑tice-based zk-SNARKs,including proof length and common reference string(CRS)*** this paper,we develop the framework of square span program-based SNARKs and design new zk-SNARKs over cyclotomic *** with previous works,our construction is without parallel repetition and achieves shorter proof and CRS lengths than previous lattice-based zk-SNARK ***,the proof length of our scheme is around 23.3%smaller than the recent shortest lattice-based zk-SNARKs by Ishai et al.(in:Proceedings of the 2021 ACM SIGSAC conference on computer and communications security,pp 212-234,2021),and the CRS length is 3.6×*** constructions follow the framework of Gennaro et al.(in:Proceedings of the 2018 ACM SIGSAC conference on computer and com‑munications security,pp 556-573,2018),and adapt it to the ring setting by slightly modifying the knowledge *** develop concretely small constructions by using module-switching and key-switching procedures in a novel way.

关键词： Zk-SNARKs Post-quantum Succinct argument

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：