Biblio

Multi-label image classification task aims to predict multiple object labels in a given image and faces the challenge of variable-sized objects. Limited by the size of CNN convolution kernels, existing CNN-based methods have difficulty capturing global dependencies and effectively fusing multiple layers features, which is critical for this task. Recently, transformers have utilized multi-head attention to extract feature with long range dependencies. Inspired by this, this paper proposes a Cross-layer Aggregation with Transformers (CAT) framework, which leverages transformers to capture the long range dependencies of CNN-based features with Long Range Dependencies module and aggregate the features layer by layer with Cross-Layer Fusion module. To make the framework efficient, a multi-head pre-max attention is designed to reduce the computation cost when fusing the high-resolution features of lower-layers. On two widely-used benchmarks (i.e., VOC2007 and MS-COCO), CAT provides a stable improvement over the baseline and produces a competitive performance.

Social networks are considered to be heterogeneous graph neural networks (HGNNs) with deep learning technological advances. HGNNs, compared to homogeneous data, absorb various aspects of information about individuals in the training stage. That means more information has been covered in the learning result, especially sensitive information. However, the privacy-preserving methods on homogeneous graphs only preserve the same type of node attributes or relationships, which cannot effectively work on heterogeneous graphs due to the complexity. To address this issue, we propose a novel heterogeneous graph neural network privacy-preserving method based on a differential privacy mechanism named HeteDP, which provides a double guarantee on graph features and topology. In particular, we first define a new attack scheme to reveal privacy leakage in the heterogeneous graphs. Specifically, we design a two-stage pipeline framework, which includes the privacy-preserving feature encoder and the heterogeneous link reconstructor with gradients perturbation based on differential privacy to tolerate data diversity and against the attack. To better control the noise and promote model performance, we utilize a bi-level optimization pattern to allocate a suitable privacy budget for the above two modules. Our experiments on four public benchmarks show that the HeteDP method is equipped to resist heterogeneous graph privacy leakage with admirable model generalization.

Intrusion detection for Controller Area Network (CAN) protocol requires modern methods in order to compete with other electrical architectures. Fingerprint Intrusion Detection Systems (IDS) provide a promising new approach to solve this problem. By characterizing network traffic from known ECUs, hazardous messages can be discriminated. In this article, a modified version of Fingerprint IDS is employed utilizing both step response and spectral characterization of network traffic via neural network training. With the addition of feature set reduction and hyperparameter tuning, this method accomplishes a 99.4% detection rate of trusted ECU traffic.

Intelligent Virtual Agents (IVAs) have become ubiquitous in our daily lives, displaying increased complexity of form and function. Initial IVA development efforts provided basic functionality to suit users' needs, typically in work or educational settings, but are now present in numerous contexts in more realistic, complex forms. In this paper, we focus on personalization of embodied human intelligent virtual agents to assist individuals as part of physical training “exergames”.

Modern cyber-physical systems that comprise controlled power electronics are becoming more internet-of-things-enabled and vulnerable to cyber-attacks. Therefore, hardening those systems against cyber-attacks becomes an emerging need. In this paper, a model-based deep learning cyber-attack detection to protect electric drive systems from cyber-attacks on the physical level is proposed. The approach combines the model physics with a deep learning-based classifier. The combination of model-based and deep learning will enable more accurate cyber-attack detection results. The proposed cyber-attack detector will be trained and simulated on a PM based electric drive system to detect false data injection attacks on the drive controller command and sensor signals.

Electrical substations in power grid act as the critical interface points for the transmission and distribution networks. Over the years, digital technology has been integrated into the substations for remote control and automation. As a result, substations are more prone to cyber attacks and exposed to digital vulnerabilities. One of the notable cyber attack vectors is the malicious command injection, which can lead to shutting down of substations and subsequently power outages as demonstrated in Ukraine Power Plant Attack in 2015. Prevailing measures based on cyber rules (e.g., firewalls and intrusion detection systems) are often inadequate to detect advanced and stealthy attacks that use legitimate-looking measurements or control messages to cause physical damage. Additionally, defenses that use physics-based approaches (e.g., power flow simulation, state estimation, etc.) to detect malicious commands suffer from high latency. Machine learning serves as a potential solution in detecting command injection attacks with high accuracy and low latency. However, sufficient datasets are not readily available to train and evaluate the machine learning models. In this paper, focusing on this particular challenge, we discuss various approaches for the generation of synthetic data that can be used to train the machine learning models. Further, we evaluate the models trained with the synthetic data against attack datasets that simulates malicious commands injections with different levels of sophistication. Our findings show that synthetic data generated with some level of power grid domain knowledge helps train robust machine learning models against different types of attacks.

Information security construction is a social issue, and the most urgent task is to do an excellent job in information risk assessment. The bayesian neural network currently plays a vital role in enterprise information security risk assessment, which overcomes the subjective defects of traditional assessment results and operates efficiently. The risk quantification method based on fuzzy theory and Bayesian regularization BP neural network mainly uses fuzzy theory to process the original data and uses the processed data as the input value of the neural network, which can effectively reduce the ambiguity of language description. At the same time, special neural network training is carried out for the confusion that the neural network is easy to fall into the optimal local problem. Finally, the risk is verified and quantified through experimental simulation. This paper mainly discusses the problem of enterprise information security risk assessment based on a Bayesian neural network, hoping to provide strong technical support for enterprises and organizations to carry out risk rectification plans. Therefore, the above method provides a new information security risk assessment idea.

Online information security labs intended for training and facilitating hands-on learning for distance students at master’s level are not easy to develop and administer. This research focuses on analyzing the results of a DSR project for design, development, and implementation of an InfoSec lab. This research work contributes to the existing research by putting forth an initial outline of a generalized model for design theory for InfoSec labs aimed at hands-on education of students in the field of information security. The anatomy of design theory framework is used to analyze the necessary components of the anticipated design theory for InfoSec labs in future.

Aiming at the single hopping strategy in the terminal information hopping active defense technology, a variety of heterogeneous hopping modes are introduced into the terminal information hopping system, the definition of the terminal information is expanded, and the adaptive adjustment of the hopping strategy is given. A network adversarial training simulation system is researched and designed, and related subsystems are discussed from the perspective of key technologies and their implementation, including interactive adversarial training simulation system, adversarial training simulation support software system, adversarial training simulation evaluation system and adversarial training Mock Repository. The system can provide a good environment for network confrontation theory research and network confrontation training simulation, which is of great significance.

The compressed sensing (CS) method can reconstruct images with a small amount of under-sampling data, which is an effective method for fast magnetic resonance imaging (MRI). As the traditional optimization-based models for MRI suffered from non-adaptive sampling and shallow” representation ability, they were unable to characterize the rich patterns in MRI data. In this paper, we propose a CS MRI method based on iterative shrinkage threshold algorithm (ISTA) and adaptive sparse sampling, called DSLS-ISTA-Net. Corresponding to the sampling and reconstruction of the CS method, the network framework includes two folders: the sampling sub-network and the improved ISTA reconstruction sub-network which are coordinated with each other through end-to-end training in an unsupervised way. The sampling sub-network and ISTA reconstruction sub-network are responsible for the implementation of adaptive sparse sampling and deep sparse representation respectively. In the testing phase, we investigate different modules and parameters in the network structure, and perform extensive experiments on MR images at different sampling rates to obtain the optimal network. Due to the combination of the advantages of the model-based method and the deep learning-based method in this method, and taking both adaptive sampling and deep sparse representation into account, the proposed networks significantly improve the reconstruction performance compared to the art-of-state CS-MRI approaches.

Vulnerability discovery is an important field of computer security research and development today. Because most of the current vulnerability discovery methods require large-scale manual auditing, and the code parsing process is cumbersome and time-consuming, the vulnerability discovery effect is reduced. Therefore, for the uncertainty of vulnerability discovery itself, it is the most basic tool design principle that auxiliary security analysts cannot completely replace them. The purpose of this paper is to study the source code vulnerability discovery method based on graph neural network. This paper analyzes the three processes of data preparation, source code vulnerability mining and security assurance of the source code vulnerability mining method, and also analyzes the suspiciousness and particularity of the experimental results. The empirical analysis results show that the types of traditional source code vulnerability mining methods become more concise and convenient after using graph neural network technology, and we conducted a survey and found that more than 82% of people felt that the design source code vulnerability mining method used When it comes to graph neural networks, it is found that the design efficiency has become higher.

Neural program embeddings have demonstrated considerable promise in a range of program analysis tasks, including clone identification, program repair, code completion, and program synthesis. However, most existing methods generate neural program embeddings di-rectly from the program source codes, by learning from features such as tokens, abstract syntax trees, and control flow graphs. This paper takes a fresh look at how to improve program embed-dings by leveraging compiler intermediate representation (IR). We first demonstrate simple yet highly effective methods for enhancing embedding quality by training embedding models alongside source code and LLVM IR generated by default optimization levels (e.g., -02). We then introduce IRGEN, a framework based on genetic algorithms (GA), to identify (near-)optimal sequences of optimization flags that can significantly improve embedding quality. We use IRGEN to find optimal sequences of LLVM optimization flags by performing GA on source code datasets. We then extend a popular code embedding model, CodeCMR, by adding a new objective based on triplet loss to enable a joint learning over source code and LLVM IR. We benchmark the quality of embedding using a rep-resentative downstream application, code clone detection. When CodeCMR was trained with source code and LLVM IRs optimized by findings of IRGEN, the embedding quality was significantly im-proved, outperforming the state-of-the-art model, CodeBERT, which was trained only with source code. Our augmented CodeCMR also outperformed CodeCMR trained over source code and IR optimized with default optimization levels. We investigate the properties of optimization flags that increase embedding quality, demonstrate IRGEN's generalization in boosting other embedding models, and establish IRGEN's use in settings with extremely limited training data. Our research and findings demonstrate that a straightforward addition to modern neural code embedding models can provide a highly effective enhancement.

Cyber security is everybody's responsibility. It is the capability of the person to protect or secure the use of cyberspace from cyber-attacks. Cyber security awareness is the combination of both knowing and doing to safeguard one's personal information or assets. Online threats continue to rise in the Philippines which is the focus of this study, to identify the level of cyber security awareness among the students and teachers of Occidental Mindoro State College (OMSC) Philippines. Results shows that the level of cyber security awareness in terms of Knowledge, majority of the students and teachers got the passing score and above however there are almost fifty percent got below the passing score. In terms of Practices, both the teachers and the students need to strengthen the awareness of system and browser updates to boost the security level of the devices used. More than half of the IT students are aware of the basic cyber security protocol but there is a big percentage in the Non-IT students which is to be considered. Majority of the teachers are aware of the basic cyber security protocols however the remaining number must be looked into. There is a need to intensity the awareness of the students in the proper etiquette in using the social media. Boost the basic cyber security awareness training to all students and teachers to avoid cybercrime victims.

CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart) is a widely used technology to distinguish real users and automated users such as bots. However, the advance of AI technologies weakens many CAPTCHA tests and can induce security concerns. In this paper, we propose a user-friendly text-based CAPTCHA generation method named Robust Text CAPTCHA (RTC). At the first stage, the foregrounds and backgrounds are constructed with font and background images respectively sampled from font and image libraries, and they are then synthesized into identifiable pseudo adversarial CAPTCHAs. At the second stage, we utilize a highly transferable adversarial attack designed for text CAPTCHAs to better obstruct CAPTCHA solvers. Our experiments cover comprehensive models including shallow models such as KNN, SVM and random forest, as well as various deep neural networks and OCR models. Experiments show that our CAPTCHAs have a failure rate lower than one millionth in general and high usability. They are also robust against various defensive techniques that attackers may employ, including adversarially trained CAPTCHA solvers and solvers trained with collected RTCs using manual annotation. Codes available at https://github.com/RulinShao/RTC.

This paper presents CaptchaGG, a model for recognizing linear graphical CAPTCHAs. As in the previous society, CAPTCHA is becoming more and more complex, but in some scenarios, complex CAPTCHA is not needed, and usually, linear graphical CAPTCHA can meet the corresponding functional scenarios, such as message boards of websites and registration of accounts with low security. The scheme is based on convolutional neural networks for feature extraction of CAPTCHAs, recurrent neural forests A neural network that is too complex will lead to problems such as difficulty in training and gradient disappearance, and too simple will lead to underfitting of the model. For the single problem of linear graphical CAPTCHA recognition, the model which has a simple architecture, extracting features by convolutional neural network, sequence modeling by recurrent neural network, and finally classification and recognition, can achieve an accuracy of 96% or more recognition at a lower complexity.

CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart) is an important security technique designed to deter bots from abusing software systems, which has broader applications in cyberspace. CAPTCHAs come in a variety of forms, including the deciphering of obfuscated text, transcribing of audio messages, and tracking mouse movement, among others. This paper focuses on using deep learning techniques to recognize text-based CAPTCHAs. In particular, our work focuses on generating training datasets using different CAPTCHA schemes, along with a pre-processing technique allowing for character-based recognition. We have encapsulated the CRABI (CAPTCHA Recognition with Attached Binary Images) framework to give an image multiple labels for improvement in feature extraction. Using real-world datasets, performance evaluations are conducted to validate the efficacy of our proposed approach on several neural network architectures (e.g., custom CNN architecture, VGG16, ResNet50, and MobileNet). The experimental results confirm that over 90% accuracy can be achieved on most models.

When storing face biometric samples in accordance with ISO/IEC 19794 as JPEG2000 encoded images, it is necessary to encrypt them for the sake of users’ privacy. Literature suggests selective encryption of JPEG2000 images as fast and efficient method for encryption, the trade-off is that some information is left in plaintext. This could be used by an attacker, in case the encrypted biometric samples are leaked. In this work, we will attempt to utilize a convolutional neural network to perform cryptanalysis of the encryption scheme. That is, we want to assess if there is any information left in plaintext in the selectively encrypted face images which can be used to identify the person. The chosen approach is to train CNNs for biometric face recognition not only with plaintext face samples but additionally conduct a refinement training with partially encrypted data. If this system can successfully utilize encrypted face samples for biometric matching, we can show that the information left in encrypted biometric face samples is information actually usable for biometric recognition.The method works and we can show that a supposedly secure biometric sample still contains identifying information on average over the whole database.

In recent years, radar automatic target recognition (RATR) technology based on high-resolution range profile (HRRP) has received extensive attention in various fields. However, insufficient data on non-cooperative targets seriously affects recognition performance of this technique. For HRRP target recognition under few-shot condition, we proposed a novel gaussian deep belief network based on model-agnostic meta-learning (GDBN-MAML). In the proposed method, GDBN allowed real-value data to be transmitted over the entire network, which effectively avoided feature loss due to binarization requirements of conventional deep belief network (DBN) for data. In addition, we optimized the initial parameters of GDBN by multi-task learning based on MAML. In this way, the number of training samples required by the model for new recognition tasks could be reduced. We applied the proposed method to the HRRP recognition experiments of 3 types of 3D simulated aircraft models. The experimental results showed that the proposed method had higher recognition accuracy and generalization performance under few-shot condition compared with conventional deep learning methods.

Aim: Object Detection is one of the latest topics in today’s world for detection of real time objects using Deep Belief Networks. Methods & Materials: Real-Time Object Detection is performed using Deep Belief Networks (N=24) over Convolutional Neural Networks (N=24) with the split size of training and testing dataset 70% and 30% respectively. Results: Deep Belief Networks has significantly better accuracy (81.2%) compared to Convolutional Neural Networks (47.7%) and attained significance value of p = 0.083. Conclusion: Deep Belief Networks achieved significantly better object detection than Convolutional Neural Networks for identifying real-time objects in traffic surveillance.

Higher education management has problems producing 100% of graduates capable of responding to the needs of industry while industry also is struggling to find qualified graduates that responded to their needs in part because of the inefficient way of evaluating problems, as well as because of weaknesses in the evaluation of problem-solving capabilities. The objective of this paper is to propose an appropriate classification model to be used for predicting and evaluating the attributes of the data set of the student in order to meet the selection criteria required by the industries in the academic field. The dataset required for this analysis was obtained from a private firm and the execution was carried out using Chimp Optimization Algorithm (COA) based Deep Belief Neural Network (COA-DBNN) and the obtained results are compared with various classifiers such as Logistic Regression (LR), Decision Tree (DT) and Random Forest (RF). The proposed model outperforms other classifiers in terms of various performance metrics. This critical analysis will help the college management to make a better long-term plan for producing graduates who are skilled, knowledgeable and fulfill the industry needs as well.

The selection of distribution network faults is of great significance to accurately identify the fault location, quickly restore power and improve the reliability of power supply. This paper mainly studies the fault phase selection method of distribution network based on wavelet singular entropy and deep belief network (DBN). Firstly, the basic principles of wavelet singular entropy and DBN are analyzed, and on this basis, the DBN model of distribution network fault phase selection is proposed. Firstly, the transient fault current data of the distribution network is processed to obtain the wavelet singular entropy of the three phases, which is used as the input of the fault phase selection model; then the DBN network is improved, and an artificial neural network (ANN) is introduced to make it a fault Select the phase classifier, and specify the output label; finally, use Simulink to build a simulation model of the IEEE33 node distribution network system, obtain a large amount of data of various fault types, generate a training sample library and a test sample library, and analyze the neural network. The adjustment of the structure and the training of the parameters complete the construction of the DBN model for the fault phase selection of the distribution network.

Recommenders are central in many applications today. The most effective recommendation schemes, such as those based on collaborative filtering (CF), exploit similarities between user profiles to make recommendations, but potentially expose private data. Federated learning and decentralized learning systems address this by letting the data stay on user's machines to preserve privacy: each user performs the training on local data and only the model parameters are shared. However, sharing the model parameters across the network may still yield privacy breaches. In this paper, we present Rex, the first enclave-based decentralized CF recommender. Rex exploits Trusted execution environments (TEE), such as Intel software guard extensions (SGX), that provide shielded environments within the processor to improve convergence while preserving privacy. Firstly, Rex enables raw data sharing, which ultimately speeds up convergence and reduces the network load. Secondly, Rex fully preserves privacy. We analyze the impact of raw data sharing in both deep neural network (DNN) and matrix factorization (MF) recommenders and showcase the benefits of trusted environments in a full-fledged implementation of Rex. Our experimental results demonstrate that through raw data sharing, Rex significantly decreases the training time by 18.3 x and the network load by 2 orders of magnitude over standard decentralized approaches that share only parameters, while fully protecting privacy by leveraging trustworthy hardware enclaves with very little overhead.

A recommender system aims to suggest the most relevant items to users based on their personal data. However, data privacy is a growing concern for anyone. Secure recommender system is a research direction to preserve user privacy while maintaining as high performance as possible. The most recent strategy is to use Federated Learning, a machine learning technique for privacy-preserving distributed training. In Federated Learning, a subset of users will be selected for training model using data at local systems, the server will securely aggregate the computing result from local models to generate a global model, finally that model will give recommendations to users. In this paper, we present a novel algorithm to train Collaborative Filtering recommender system specialized for the ranking task in Federated Learning setting, where the goal is to protect user interaction information (i.e., implicit feedback). Specifically, with the help of the algorithm, the recommender system will be trained by Neural Collaborative Filtering, one of the state-of-the-art matrix factorization methods and Bayesian Personalized Ranking, the most common pairwise approach. In contrast to existing approaches which protect user privacy by requiring users to download/upload the information associated with all interactions that they can possibly interact with in order to perform training, the algorithm can protect user privacy at low communication cost, where users only need to obtain/transfer the information related to a small number of interactions per training iteration. Above all, through extensive experiments, the algorithm has demonstrated to utilize user data more efficient than the most recent research called FedeRank, while ensuring that user privacy is still preserved.

Cybersecurity is important in the field of information technology. One most recent pressing issue is information security. When we think of cybersecurity, the first thing that comes to mind is cyber-attacks, which are on the rise, such as Ransomware. Various governments and businesses take a variety of measures to combat cybercrime. People are still concerned about ransomware, despite numerous cybersecurity precautions. In ransomware, the attacker encrypts the victim’s files/data and demands payment to unlock the data. Cybersecurity is a collection of tools, regulations, security guards, security ideas, guidelines, risk management, activities, training, insurance, best practices, and technology used to secure the cyber environment, organization, and user assets. This paper analyses ransomware attacks, techniques for dealing with these attacks, and future challenges.

Amount and variety of training data drastically affect the performance of CNNs. Thus, annotation methods are becoming more and more critical to collect data efficiently. In this paper, we propose a simple yet efficient Interactive Self-Annotation framework to cut down both time and human labor cost for video object bounding box annotation. Our method is based on recurrent self-supervised learning and consists of two processes: automatic process and interactive process, where the automatic process aims to build a supported detector to speed up the interactive process. In the Automatic Recurrent Annotation, we let an off-the-shelf detector watch unlabeled videos repeatedly to reinforce itself automatically. At each iteration, we utilize the trained model from the previous iteration to generate better pseudo ground-truth bounding boxes than those at the previous iteration, recurrently improving self-supervised training the detector. In the Interactive Recurrent Annotation, we tackle the human-in-the-loop annotation scenario where the detector receives feedback from the human annotator. To this end, we propose a novel Hierarchical Correction module, where the annotated frame-distance binarizedly decreases at each time step, to utilize the strength of CNN for neighbor frames. Experimental results on various video datasets demonstrate the advantages of the proposed framework in generating high-quality annotations while reducing annotation time and human labor costs.