Deepfake technologies leverage deep learning to generate highly realistic videos involving face swapping and expression transfer, often exceeding the threshold of human visual perception. This poses serious challenges to social governance and digital security, highlighting the urgent need for reliable forgery detection methods. Training detection models without using real forgeries is considered a promising strategy to improve generalization. These approaches simulate diverse forgery traces to generate synthetic training data. However, most existing methods rely on 2 D image manipulation and fail to capture 3D forgery characteristics such as geometric distortion, expression mismatch, and texture anomalies-leading to poor performance on reconstruction-based forgeries. To address this problem, we propose a Reconstruction-Blended Image (RBI) generation method based on 3D Morphable Models (3DMM). By perturbing facial shape and expression parameters, this approach produces training samples that better reflect 3D reconstruction artifacts. When combined with traditional Self-Blended Images (SBI), the hybrid training strategy enhances the model's ability to detect a wider range of forgeries. Experiments show that this method improves AUC by$\mathbf{1 0. 7 9} \boldsymbol{\%}$on challenging cases like Face2Face forgeries. In summary, our 3D face reconstruction-based generation strategy significantly enhances the generalization and robustness of forgery detection models, offering a practical solution to emerging deepfake threats.

Sophisticated and realistic facial manipulation videos created by deepfake technology have become ubiquitous, leading to profound trust crises and security risks in contemporary society. However, various researchers concentrate on enhancing the precision and generalization of deepfake detection models, with little attention to forgery localization. Detecting deepfakes and identifying fake regions is a challenging task. We propose an end-to-end model for performing deepfake detection and forgery localization based on the Laplacian pyramid. The model is designed by an encoder–decoder architecture. Specifically, the encoder generates multi-scale features. The decoder gradually integrates multi-scale features and Laplacian residuals to reconstruct the prediction masks coarse-to-finely. Otherwise, we adopt a spatial pyramid pool approach to deal with high-level semantic features and integrate local and global information. Comprehensive experiments demonstrate that the proposed model performs satisfactorily in deepfake detection and localization.

The widespread adoption of face manipulation systems has brought entertainment and convenience to users while posing significant challenges to media forensics. Conventional active defense strategies typically generate adversarial images by introducing perturbations into the original images. When adversarial images undergo facial manipulation, they often exhibit distortions or speckle artifacts, which helps reduce the dissemination of forged content on social media platforms. Nevertheless, the widespread dissemination of degraded images may contribute to facial stigmatization. Furthermore, conventional perturbation techniques are vulnerable to failure under JPEG compression and various image processing operations on OSN platforms. To address these challenges, we introduce a robust and unstigmatized imperceptible perturbation (RUIP) method designed to counteract face manipulation. First, RUIP utilizes an end-to-end adversarial training framework to generate robust and imperceptible perturbations. Second, to mitigate facial stigmatization, we incorporate both pixel-level and feature-level guidance losses during training, ensuring that the output images remain visually natural and closely aligned with the original images. Finally, we develop a novel module, the Flexible Random Enhancement Generator (FREG), to simulate complex JPEG compression and diverse image processing operations on OSN platforms, enhancing the model’s robustness against perturbations. Extensive qualitative and quantitative experiments demonstrate that the proposed method effectively defends against face manipulation attacks while preserving the visual quality of facial images under JPEG compression and other image processing operations on OSN platforms. We propose an effective and unstigmatized defense algorithm to safeguard privacy and maintain the stability of the social media ecosystem.Code is available athttps://github.com/silencecmsj/RUIP.

Face forgery videos, known as Deepfakes, are widely spread on social media with great potential threat, making the detection of forged face videos is crucial. Commonly, forgery video detection methods are based on Convolutional Neural Networks or Transformer, which treats images as grid or sequence structures for binary classification discrimination. Since face objects are usually not regularly shaped quadrilaterals, treating them as grid or sequence structures is redundant and inflexible, thus losing useful information. Based on this, we propose a new perspective to represent facial images as graph structures, which are fed into Graph Neural Network to learn the intrinsic relationships of facial regions for deep forgery detection. In addition, we propose a feature fusion module to learn artifact information in the frequency domain for a more comprehensive facial feature representation to further improve the reliability of our model. Extensive experiments on several benchmark databases demonstrate the effectiveness and robust generalization ability of our method compared with many state-of-the-art methods.

As the demand for the Internet of Things (IoT) grows, it becomes crucial to possess systems capable of detecting any data leakage used for authentication. Within IoT camera systems based on facial bio-metric recognition, there is a risk of Deepfake Bypassed Facial Feature Authentication due to the widespread use of deepfake video technologies, such as DeepFaceLive and expression manipulation. Traditional Face Anti-Spoofing Detection techniques may struggle to detect real-time deepfake videos within IoT contexts. Moreover, constrained by the scale of Face Anti-Spoofing Detection datasets, current detection models primarily focus on recognizing the entire face in videos, neglecting the inter-component correlations of facial features. However, our investigation indicates that different parts of the face have varying impacts on deepfake detection. To address this issue, we segment the face into several regions within video frames and explore the relationships between these regions. Our approach involves constructing feature graphs that represent such correlations, aiming to leverage the relationships between facial regions and the temporal characteristics of real-time facial manipulation videos for use in live facial detection cameras. Initially, features for each facial region are extracted via Convolutional Neural Networks (CNNs). Subsequently, with these features as vertices and their correlations as edges, a feature graph of the entire video is constructed. Ultimately, a Graph Neural Network (GNN) is employed to determine whether the video has been tampered with. Experiments conducted on several publicly accessible datasets demonstrate that our proposed method outperforms other state-of-the-art Face Anti-Spoofing Detection techniques in most scenarios. Thus, the aforementioned advanced Graph Neural Network model exhibits exceptional performance in real-time deepfake detection tailored for live facial detection cameras.

With the continuous advancement of deepfake technology, there has been a surge in the creation of realistic fake videos. Unfortunately, the malicious utilization of deepfake poses a significant threat to societal morality and political security. Therefore, numerous researchers have proposed various deepfake detection methods. However, traditional deepfake approaches tend to focus on specific forgery features, such as artifacts or inconsistent actions, which can be vulnerable to specialized countermeasures. Recent studies show an intrinsic correlation between facial and audio cues, which can be exploited for deepfake detection. To address these challenges and enhance the robustness and generalization of deepfake detection algorithms, we propose a novel joint audio-visual deepfake detection model named AVA-CL, which is capable of detecting deepfakes in both audio and visual domains. Furthermore, exploiting the inherent correlation and consistency between audio and visual enhances the effectiveness of deepfake detection significantly. Through extensive experiments, we demonstrate that our proposed AVA-CL model outperforms many state-of-the-art (SOTA) methods with superior robustness and generalization capabilities. This research presents a promising approach for deepfake detection and reducing the harm caused by malicious use.

With the increasing development of deepfake in facial editing and the easy accessibility of image and video content on the internet, the security risks of spreading false personal information through social media platforms are becoming increasingly serious. The harm caused by deepfake includes spreading false information, pornography, financial fraud, privacy breaches, and security system damage. To address these security issues, it is necessary to strengthen the detection and defense of deepfake. Current research mainly focuses on the detection of deepfake images and videos. In recent years, some work has proposed using the method of generating adversarial samples to deal with the malicious operations of GAN networks in deepfake. And it has proposed different single perturbation fusion methods to generate universal adversarial watermarks that can defend against modifications by multiple models. However, all of these works generate single-step watermarks using gradient-based methods, which cannot accurately control the required perturbation strength, resulting in large errors and more irrelevant information compared to the original image after superimposing the watermark. To solve this problem, we propose an optimized adversarial face deepfake watermark and measure the protection success rate and defense performance of this method on single and multiple models. From the extensive experimental results, we find that the perturbation generated by the Optimization-based method can successfully generate smaller perturbations while ensuring a high protection success rate, making the adversarial samples closer to the original samples. It can also be applied to different types of models and loss functions, accurately generating perturbation strength, and has a wider range of applicability.

Edge AI technologies have been used for many Intelligent Transportation Systems, such as road traffic monitor systems. Neural Architecture Search (NAS) is a typcial way to search high-performance models for edge devices with limited computing resources. However, NAS is also vulnerable to adversarial attacks. In this paper, A One-Shot NAS is employed to realize derivative models with different scales. In order to study the relation between adversarial robustness and model scales, a graph-based method is designed to select best sub models generated from One-Shot NAS. Besides, an evaluation method is proposed to assess robustness of deep learning models under various scales of models. Experimental results shows an interesting phenomenon about the correlations between network sizes and model robustness, reducing model parameters will increase model robustness under maximum adversarial attacks, while, increasing model paremters will increase model robustness under minimum adversarial attacks. The phenomenon is analyzed, that is able to help understand the adversarial robustness of models with different scales for edge AI transportation systems.

The cloud computing, as a more distributed and more efficient paradigm with better performance, has played an important role in many fields. The cloud environment under the IoT has also assumed many roles such as the storage, the processor, the service provider and so on. However, the complex deployment and usage environment of the IoT brings new security risks to cloud computing. In response to this situation, this poster preliminarily designed a security scheme for the cloud environment of IoT. Based on the identity verification algorithms and blockchain technology, the credibility of data stored in the cloud can be ensured, while the security of data transmission from the cloud to data consumers is achieved.

Non-intrusive detection is critical to protecting the integrity of pipelines. Based on the wireless sensor network, a novel leak monitoring system, composed of a computer center, a coordinator and wireless non-intrusive sensing nodes, is proposed for pipelines of closed spaces in this paper. The wireless nonintrusive sensing node with convenient installation and disassembly on the pipeline wall is designed. The proposed system can achieve signal synchronous sampling of all wireless non-intrusive sensing nodes by the coordinator wirelessly broadcasting the time information from its GPS to them, which is significant to guarantee the accuracy of the leak location. Based on the delay cross-correlation analysis, a leak location method is presented for multiple sensors. And experimental results demonstrate that the proposed system can accurately detect and locate pipeline leaks.

There are lots of different sample points in a single trace, the each sample point containing some leakage information is useful to obtain the key when a chip encrypt plaintext with the key, these points of interest in a trace could be extracted and a new trace is formed sequentially. If using this shorter trace could improve the performance of classification in neural networks during side-channel attacks and to reduce the amount of traces required in the classification, and it means that these sample points are indeed useful and contain lots of information needed for side-channel attacks. In this paper, different amount of points of interest extracted from traces in ASCAD to form a new kind of traces as input data feed into neural networks including Multi-Layer Perceptron, Convolutional Neural Networks. In order to compare with the result of POI-traces, Principal Components Analysis was also used to shorter the length of the original trace in ASCAD, so that its length is the same as POI-traces'. About the results, the classification results with ASCAD traces are the worst using MLP or CNN, the results of PCA-traces which contain only 100 sample points using MLP are the best, and the results of POI-traces containing 300 sample points are the best result using CNN. So when using neural networks to assist side-channel attacks, the transformation of traces or the reduction of its length is advantageous to a certain extent.

In order to cooperate with the formulation of the "People's Republic of China Radio and Television Industry Standards-- Technical specification of digital rights management for video audio content distribution", a fusion media copyright protection product and system test evaluation platform supporting domestic passwords was constructed, and the end-to-end multimedia copyright protection system was implemented. Test evaluation requires compatibility testing of the DRM (Digital Rights Management) system. This paper captures and filters the HTTP protocol packets in the 4K-DRM system, and then analyzes the compliance of the HTTP packet protocol such as key synchronization, key query, and content authorization to determine whether the system passes the compatibility test.

As an organization that develops DRM (digital right management) standards in China, ChinaDRM proposed a set of solutions for the DRM application. This paper mainly focuses on the rights expression and the content authorization mechanism of ChinaDRM. We built a basic model for ChinaDRM's rights expression, introduced the format of ChinaDRM license and gave a license instance for analysis. At the end of this paper, we introduced the content authorization mechanism of ChinaDRM.

Medical insurance management is the core content of social security management. In recent years, there are many events about medical insurance illegal behaviors cause a large number of medical insurance fund losses. How to effectively supervise the medical insurance fund to protect the legal rights of the insured person has become an urgent problem. Due to the low efficiency of audit work and the hidden illegal behavior, traditional medical insurance supervision systems often waste massive resources, but still cannot play an effective role. Based on analyzing the status of medical insurance supervision and the importance of transforming supervision mode, this paper mainly studies the design and implementation of an intelligent medical insurance supervision system.

Due to the increasing demand for domestic digital audio and video services, broadcast TV intelligent terminals are rapidly iterating, and global content producers are turning to researching 4K HDR technology. China needs to realize the importance of digital rights, which is more conducive to the development of ultra-high-definition digital services in China. Therefore, this paper will introduce the architecture of China digital rights management (ChinaDRM) and its application in TS encapsulation format. In the meantime, we proposed an analytical method about the TS bitstream which had been encrypted. In the TS bitstream, it also carried the ChinaDRM descriptor and Content Encryption Information (CEI). Finally, we designed a bitstream parsing system to parse the data of CEI and ChinaDRM descriptor.

Big data is increasingly becoming a hot research topic, applied to all walks of life. And the film and television big data makes big data science and film and television industry blend together, making far-reaching impact on the film and television works of creation, dissemination, acceptance and other aspects. With the hit of different types of films and TV plays, the screenwriter directing industry has become a lot of people dream career, they eager to show talent in this area. The significance of this paper is to use the online writing mode of the script to draw the figure of character relationship. It can clearly grasp the characters in the script by visualization, and make the decision support of the characters and balance the role relation to ensure the scriptwriter's layout is reasonable. Based on the current situation of script creation and the actual needs of users, so that different levels of users can freely create character relationship map, and give the role of quantitative analysis of the results, so as to achieve the purpose of supporting the creation.

Stacking model Fusion is a combination classification method for natural language processing and text categorization. Compared to a single weak classifier, model Fusion has the advantage of combining the classification strengths of multiple classifiers, so the combination classifier is often more accurate than a single classifier, and the research of this field has been developed rapidly in recent years, and the combination classifier has been applied in various natural language processing tasks. But only by using the prediction results of the first layer weak classifier to train the second layer classifier, it has a strong limitation, only considers the training of the classification result and ignores the semantic information. We think that the method of training the weak classifier by TFIDF to the document, the expression of the document is not enough, only the information about the frequency of the document and the document is lack of the semantic information of the word2vector. In this paper, a new combination classification method is proposed, which combines the various weak classifiers trained by TFIDF and Word2vector to express the documents in many aspects, and the feature expression can fully utilize the information provided by the document. It has better classification effect than individual word2vector expression and classification and simple weak classifier combination classification.

With the rapid development of the Internet, there are more and more ways of spreading news, and it's spreading faster and faster. This creates a lot of fake news that confuses the reader's vision. In order to construct a healthy news communication environment, it is necessary to suppress fake news and to crack down on the source of fake news, so it is necessary to trace the source of news. This paper tracks the news based on the distributed storage, decentralization and other features of the blockchain, with the technology of consensus algorithm and intelligent contract.

This paper studies the development history and research status of text mining. Text mining technology commonly used to obtain valuable information and knowledge. Based on the use of open source Python modules , " Art of War of Sun Tzu " was analyzed through the visualization technology to show the key contents of the book, so that everyone has a more vivid and intuitive understanding of this famous masterpiece.

In 1976, S. Even and Cooper proved that Course Timetabling Problem (CTP) is a nondeterministic polynomial time-complete problem which is hard to be solved. Though many intelligent methods have been applied to solve the Course Timetabling Problem in recent years, there is still a lot of room for improvement. This paper proposed a Greedy and Genetic Fusion Algorithm (GGFA) to solve Course Timetabling Problem efficiently, which can obtain the local optimal solution by using Greedy Algorithm and provide a high-quality initial population for Genetic Algorithm. Simulation results prove that Greedy and Genetic Fusion Algorithm proposed in this paper has the stronger optimal ability and faster convergence speed than standard Genetic Algorithm, which could solve the CTP effectively and achieved good results.

In the process of load forecasting under normal operation condition, there may be some of bad data in observing data of the power distributing systems, which will affect the reliability and accuracy of the processing result. Therefore, to detect and identify these bad data is particularly important. Fuzzy clustering analysis is a common method of bad data detection and recognition in power system, but its extreme sensitivity to the initial cluster center will lead to the inaccuracy of the classification results. In this paper, the power data is excavated on the basis of the hierarchical clustering algorithm, the gap statistical algorithm (GSA) and autoregressive integrated moving average model (ARMA), so as to complete bad data detection and recognition of the power system. In order to verify the correctness and effectiveness of the algorithm, the algorithm program is written in MATLAB, and the simulation analysis is carried out on the basis of massive power data in XIAMEN. The results show that the algorithm can effectively identify and reject bad data in power system, and therefore laying foundation for state estimation and medium and long term load forecasting of the power system.

This paper discusses the necessity and importance of the establishment of big data from the current predicament of cable TV industry, and introduces the current situation of the development of big data of Internet and telecommunication industry,and expounds that the cable industry how to deal with and develop big data from many aspects,which ensure that cable industry can have the dominant position in the fierce competition.

With the development of the Internet, people's production and life has undergone tremendous changes. People's way of study is no longer limited to books and words, online education is becoming a trend, more and more online education system has been developed, like mooc, coursera and so on.Seemingly simple online education platform, which has a complex architecture design, we not only need to consider the amount of concurrent,we should also consider the fluency of video play,as well as asynchronous batch operations, etc. This paper I will describes the architecture design of the server from four aspects: load balance, streaming media server, asynchronous queue, distributed storage.

“Big data”, the word has been mentioned more and more. It is better to say that we are in such an era rather than that the “big data” era is coming. How to use good data, so that the hands of the data collected into a valuable thing, rather than a lot of data garbage, is the problem that every industry has been thinking about. Every enterprise wants to seize the opportunity to develop large data, in order to walk in the forefront of the industry, thus becoming the industry leader, bring greater benefits for themselves. This article focuses on some of the existing problems of radio and television and some suggestions for the big data platform construction.

Coming with the big data era, our traditional TV media is under the impact of new media, directly reflect in the loss of customers and a serious decline in advertising. This paper analyzes the main problem television media confronting in the big data era, calls the need for the big data platform of television media to solve these problems, and proposes the general idea of big data platform television media and the big data platform framework of television media.

Cloud computing is the product of a traditional classical computer technology, communications technology integration and development, also the key technology to lead information industry means to the future of innovation. This article describes the key technology of cloud computing, reviews cloud computing for industry development status and analyzes the applications of radio and television carrier services, researches how to build a legitimate broadcasting operators cloud computing architecture for supporting the corresponding business applications, how to improve user perception, reducing stress operator core system for enterprises and make cost efficiency.

As a kind of mature algorithm, genetic algorithm has been widely used in the field of Artificial intelligence and has played an important role in promoting the development of artificial intelligence technology. This paper analyzes the principle and characteristics of genetic algorithm and introduces an improved algorithm combining with simulated annealing algorithm and genetic algorithm. As an example of solving the TSP problem, this paper analyzes the advantages of the improved algorithm. We will discuss the development direction and prospect of genetic algorithm in the field of artificial intelligence.