Biblio

Although several different attacks or modern security mechanisms have been proposed, the captchas created by the numbers and the letters are still used by some websites or applications to protect their information security. The reason is that the labels of the captcha data are difficult to collect for the attacker, and protector can easily control the various parameters of the captchas: like the noise, the font type, the font size, and the background color, then make this security mechanism update with the increased attack methods. It can against attacks in different situations very effectively. This paper presents a method to recognize the different text-based captchas based on a system constituted by the denoising autoencoder and the Convolutional Recurrent Neural Network (CRNN) model with the Connectionist Temporal Classification (CTC) structure. We show that our approach has a better performance for recognizing, and it solves the identification problem of indefinite character length captchas efficiently.

Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA) is a type of automatic program to determine whether the user is human or not. The most common type of CAPTCHA is a kind of message interpretation by twisting the letters and adding slight noises in the background, plays a role of verification code. In this paper, we will introduce the basis of Convolutional Neural Network first. Then based on the handwritten digit recognition using CNN, we will develop a network for CAPTCHA image recognition.

CAPTCHA or Completely Automated Public Turing test to Tell Computers and Humans Apart is a technique to distinguish between humans and computers by generating and evaluating tests that can be passed by humans but not computer bots. However, captchas are not foolproof, and they can be bypassed which raises security concerns. Hence, sites over the internet remain open to such vulnerabilities. This research paper identifies the vulnerabilities found in some of the commonly used captcha schemes by cracking them using Deep Learning techniques. It also aims to provide solutions to safeguard against these vulnerabilities and provides recommendations for the generation of secure captchas.

Bitcoin has received a lot of attention from investors, researchers, regulators, and the media. It is a known fact that the Bitcoin price usually fluctuates greatly. However, not enough scientific research has been done on these fluctuations. In this study, long short-term memory (LSTM) modeling from Recurrent Neural Networks, which is one of the deep learning methods, was applied on Bitcoin values. As a result of this application, anomaly detection was carried out in the values from the data set. With the LSTM network, a time-dependent representation of Bitcoin price can be captured, and anomalies can be selected. The factors that play a role in the formation of the model to be applied in the detection of anomalies with the experimental results were evaluated.

When biometric databases grow larger, a security breach or leak can affect millions. In order to protect against such a threat, the use of encryption is a natural choice. However, a biometric identification attempt then requires the decryption of a potential huge database, making a traditional approach potentially unfeasible. The use of selective JPEG2000 encryption can reduce the encryption’s computational load and enable a secure storage of biometric sample data. In this paper we will show that selective encryption of face biometric samples is secure. We analyze various encoding settings of JPEG2000, selective encryption parameters on the "Labeled Faces in the Wild" database and apply several traditional and deep learning based face recognition methods.

Enhancement of the vanilla generative adversarial network (GAN) to preserve data variability in the presence of real world noise is of paramount significance in deep learning. In this study, we proposed a new distance metric of cosine distance in the framework of optimal transport (OT), and presented and validated a convolutional neural network (CNN) based GAN framework. In comparison with state-of-the-art methods based on Graphics Processing Units (GPU), the proposed framework could maintain the data diversity and quality best in terms of inception score (IS), Fréchet inception distance (FID) and enhancing the classification network of bone age, and is robust to noise degradation. The proposed framework is independent of hardware and thus could also be extended to more advanced hardware such as specialized Tensor Processing Units (TPU), and could be a potential built-in component of a general deep learning networks for such applications as image classification, segmentation, registration, and object detection.

Malicious Webpages have been a serious threat on Internet for the past few years. As per the latest Google Transparency reports, they continue to be top ranked amongst online threats. Various techniques have been used till date to identify malicious sites, to include, Static Heuristics, Honey Clients, Machine Learning, etc. Recently, with the rapid rise of Deep Learning, an interest has aroused to explore Deep Learning techniques for detecting Malicious Webpages. In this paper Deep Learning has been utilized for such classification. The model proposed in this research has used a Deep Neural Network (DNN) with two hidden layers to distinguish between Malicious and Benign Webpages. This DNN model gave high accuracy of 99.81% with very low False Positives (FP) and False Negatives (FN), and with near real-time response on test sample. The model outperformed earlier machine learning solutions in accuracy, precision, recall and time performance metrics.

This study shows the effectiveness of anomaly-based IDS using long short-term memory(LSTM) based on the newly developed dataset called UNSW-NB15 while considering root mean square error and mean absolute error as evaluation metrics for accuracy. For each attack, 80% and 90% of samples were used as LSTM inputs and trained this model while increasing epoch values. Furthermore, this model has predicted attack points by applying test data and produced possible attack points for each attack at the 3rd time frame against the actual attack point. However, in the case of an Exploit attack, the consecutive overlapping attacks happen, there was ambiguity in the interpretation of the numerical values calculated by the LSTM. We presented a methodology for training data with binary values using LSTM and evaluation with RMSE metrics throughout this study.

Recently, as a result of the COVID-19 pandemic, the internet service has seen an upsurge in use. As a result, the usage of cloud computing apps, which offer services to end users on a subscription basis, rises in this situation. However, the availability and efficiency of cloud computing resources are impacted by DDoS attacks, which are designed to disrupt the availability and processing power of cloud computing services. Because there is no effective way for detecting or filtering DDoS attacks, they are a dependable weapon for cyber-attackers. Recently, researchers have been experimenting with machine learning (ML) methods in order to create efficient machine learning-based strategies for detecting DDoS assaults. In this context, we propose a technique for detecting DDoS attacks in a cloud computing environment using big data and deep learning algorithms. The proposed technique utilises big data spark technology to analyse a large number of incoming packets and a deep learning machine learning algorithm to filter malicious packets. The KDDCUP99 dataset was used for training and testing, and an accuracy of 99.73% was achieved.

Wireless communications networks are an integral part of intelligent systems that enhance the automation of various activities and operations embarked by humans. For example, the development of intelligent devices imbued with sensors leverages emerging technologies such as machine learning (ML) and artificial intelligence (AI), which have proven to enhance military operations through communication, control, intelligence gathering, and situational awareness. However, growing concerns in cybersecurity imply that attackers are always seeking to take advantage of the widened attack surface to launch adversarial attacks which compromise the activities of legitimate users. To address this challenge, we leverage on deep learning (DL) and the principle of cyber-deception to propose a method for defending wireless networks from the activities of jammers. Specifically, we use DL to regulate the power allocated to users and the channel they use to communicate, thereby luring jammers into attacking designated channels that are considered to guarantee maximum damage when attacked. Furthermore, by directing its energy towards the attack on a specific channel, other channels are freed up for actual transmission, ensuring secure communication. Through simulations and experiments carried out, we conclude that this approach enhances security in wireless communication systems.

Recently deep learning has demonstrated much success within the fields of image and natural language processing, facial recognition, and computer vision. The success is attributed to large, accessible databases and deep learning's ability to learn highly accurate models. Thus, deep learning is being investigated as a viable end-to-end approach to digital communications design. This work investigates the use of adversarial deep learning to ensure that a radio can communicate covertly, via Direct Sequence Spread Spectrum (DSSS), with another while a third (the adversary) is actively attempting to detect, intercept and exploit their communications. The adversary's ability to detect and exploit the DSSS signals is hindered by: (i) generating a set of spreading codes that are balanced and result in low side lobes as well as (ii) actively adapting the encoding scheme. Lastly, DSSS communications performance is assessed using energy constrained devices to accurately portray IoT and IoBT device limitations.

In the era of Internet of Things (IoT), the connection links are established from devices easily, which is vulnerable to insecure attacks from intruders, hence intrusion detection system in IoT is the need of an hour. One of the important thing for any organization is securing the confidential information and data from outside attacks as well as unauthorized access. There are many attempts made by the researchers to develop the strong intrusion detection system having high accuracy. These systems suffer from many disadvantages like unacceptable accuracy rates including high False Positive Rate (FPR) and high False Negative Rate (FNR), more execution time and failure rate. More of these system models are developed by using traditional machine learning techniques, which have performance limitations in terms of accuracy and timeliness both. These limitations can be overcome by using the deep learning techniques. Deep learning techniques have the capability to generate highly accurate results and are fault tolerant. Here, the intrusion detection model for IoT is designed by using the Taylor-Spider Monkey optimization (Taylor-SMO) which will be developed to train the Deep belief neural network (DBN) towards achieving an accurate intrusion detection model. The deep learning accuracy gets increased with increasing number of training data samples and testing data samples. The optimization based algorithm for training DBN helps to reduce the FPR and FNR in intrusion detection. The system will be implemented by using the NSL KDD dataset. Also, this model will be trained by using the samples from this dataset, before which feature extraction will be applied and only relevant set of attributes will be selected for model development. This approach can lead to better and satisfactory results in intrusion detection.

In order to improve the security and anti-interference ability of industrial control system, this paper proposes an improved industrial neural network defense method based on the PCA dimension reduction and the improved deep neural network. Firstly, the proposed method reduces the dimensionality of the industrial data using the dimension reduction theory of principal component analysis (PCA). Then the deep neural network extracts the features of the network. Finally, the softmax classifier classifies industrial data. Experiment results show that compared with unintegrated algorithm, this method achieves higher recognition accuracy and has great application potential.

Algorithms for unsupervised anomaly detection have proven their effectiveness and flexibility, however, first it is necessary to calculate with what ratio a certain class begins to be considered anomalous by the autoencoder. For this reason, we propose to conduct a study of the efficiency of autoencoders depending on the ratio of anomalous and non-anomalous classes. The emergence of high-speed networks in electric power systems creates a tight interaction of cyberinfrastructure with the physical infrastructure and makes the power system susceptible to cyber penetration and attacks. To address this problem, this paper proposes an innovative approach to develop a specification-based intrusion detection framework that leverages available information provided by components in a contemporary power system. An autoencoder is used to encode the causal relations among the available information to create patterns with temporal state transitions, which are used as features in the proposed intrusion detection. This allows the proposed method to detect anomalies and cyber attacks.

Cloud Computing is a favored choice of any IT organization in the current context since that provides flexibility and pay-per-use service to the users. Moreover, due to its open and inclusive architecture which is accessible to attackers. Security and privacy are a big roadblock to its success. For any IT organization, intrusion detection systems are essential to the detection and endurance of effective detection system against attacker aggressive attacks. To recognize minor occurrences and become significant breaches, a fully managed intrusion detection system is required. The most prevalent approach for intrusion detection on the cloud is the Intrusion Detection System (IDS). This research introduces a cloud-based deep learning-LSTM IDS model and evaluates it to a hybrid Stacked Contractive Auto Encoder (SCAE) + Support Vector Machine (SVM) IDS model. Deep learning algorithms like basic machine learning can be built to conduct attack detection and classification simultaneously. Also examine the detection methodologies used by certain existing intrusion detection systems. On two well-known Intrusion Detection datasets (KDD Cup 99 and NSL-KDD), our strategy outperforms current methods in terms of accurate detection.

Intrusion Detection System (IDS) is used to identify malicious traffic on the network. Apart from rule-based IDS, machine learning and deep learning based on IDS are also being developed to improve the accuracy of IDS detection. In this study, the public dataset CIC IDS 2017 was used in developing deep learning-based IDS because this dataset contains the new types of attacks. In addition, this dataset also meets the criteria as an intrusion detection dataset. The dataset was split into train data, validation data and test data. We proposed Bidirectional Long-Short Term Memory (LSTM) for building neural network. We created 24 scenarios with various changes in training parameters which were trained for 100 epochs. The training parameters used as research variables are optimizer, activation function, and learning rate. As addition, Dropout layer and L2-regularizer were implemented on every scenario. The result shows that the model used Adam optimizer, Tanh activation function and a learning rate of 0.0001 produced the highest accuracy compared to other scenarios. The accuracy and F1 score reached 97.7264% and 97.7516%. The best model was trained again until 1000 iterations and the performance increased to 98.3448% in accuracy and 98.3793% in F1 score. The result exceeded several previous works on the same dataset.

Ever since Machine Learning as a Service emerges as a viable business that utilizes deep learning models to generate lucrative revenue, Intellectual Property Right (IPR) has become a major concern because these deep learning models can easily be replicated, shared, and re-distributed by any unauthorized third parties. To the best of our knowledge, one of the prominent deep learning models - Generative Adversarial Networks (GANs) which has been widely used to create photorealistic image are totally unprotected despite the existence of pioneering IPR protection methodology for Convolutional Neural Networks (CNNs). This paper therefore presents a complete protection framework in both black-box and white-box settings to enforce IPR protection on GANs. Empirically, we show that the proposed method does not compromise the original GANs performance (i.e. image generation, image super-resolution, style transfer), and at the same time, it is able to withstand both removal and ambiguity attacks against embedded watermarks. Codes are available at https://github.com/dingsheng-ong/ipr-gan.

Deep Neural Networks (DNN) has gained great success in solving several challenging problems in recent years. It is well known that training a DNN model from scratch requires a lot of data and computational resources. However, using a pre-trained model directly or using it to initialize weights cost less time and often gets better results. Therefore, well pre-trained DNN models are valuable intellectual property that we should protect. In this work, we propose DeepTrace, a framework for model owners to secretly fingerprinting the target DNN model using a special trigger set and verifying from outputs. An embedded fingerprint can be extracted to uniquely identify the information of model owner and authorized users. Our framework benefits from both white-box and black-box verification, which makes it useful whether we know the model details or not. We evaluate the performance of DeepTrace on two different datasets, with different DNN architectures. Our experiment shows that, with the advantages of combining white-box and black-box verification, our framework has very little effect on model accuracy, and is robust against different model modifications. It also consumes very little computing resources when extracting fingerprint.

Internal attacks are one of the biggest cybersecurity issues to companies and businesses. Despite the implemented perimeter security systems, the risk of adversely affecting the security and privacy of the organization’s information remains very high. Actually, the detection of such a threat is known to be a very complicated problem, presenting many challenges to the research community. In this paper, we investigate the effectiveness and usefulness of using Autoencoder and Variational Autoencoder deep learning algorithms to automatically defend against insider threats, without human intervention. The performance evaluation of the proposed models is done on the public CERT dataset (CERT r4.2) that contains both benign and malicious activities generated from 1000 simulated users. The comparison results with other models show that the Variational Autoencoder neural network provides the best overall performance with a higher detection accuracy and a reasonable false positive rate.

Currently, more and more malicious insiders are making threats, and the detection of insider threats is becoming more challenging. The malicious insider often uses legitimate access privileges and mimic normal behaviors to evade detection, which is difficult to be detected via using traditional defensive solutions. In this paper, we propose DeepMIT, a malicious insider threat detection framework, which utilizes Recurrent Neural Network (RNN) to model user behaviors as time sequences and predict the probabilities of anomalies. This framework allows DeepMIT to continue learning, and the detections are made in real time, that is, the anomaly alerts are output as rapidly as data input. Also, our framework conducts further insight of the anomaly scores and provides the contributions to the scores and, thus, significantly helps the operators to understand anomaly scores and take further steps quickly(e.g. Block insider's activity). In addition, DeepMIT utilizes user-attributes (e.g. the personality of the user, the role of the user) as categorical features to identify the user's truly typical behavior, which help detect malicious insiders who mimic normal behaviors. Extensive experimental evaluations over a public insider threat dataset CERT (version 6.2) have demonstrated that DeepMIT has outperformed other existing malicious insider threat solutions.

Deep learning has undergone tremendous advancements in computer vision studies. The training of deep learning neural networks depends on a considerable amount of ground truth datasets. However, labeling ground truth data is a labor-intensive task, particularly for large-volume video analytics applications such as video surveillance and vehicles detection for autonomous driving. This paper presents a rapid and accurate method for associative searching in big image data obtained from security monitoring systems. We developed a semi-automatic moving object annotation method for improving deep learning models. The proposed method comprises three stages, namely automatic foreground object extraction, object annotation in subsequent video frames, and dataset construction using human-in-the-loop quick selection. Furthermore, the proposed method expedites dataset collection and ground truth annotation processes. In contrast to data augmentation and data generative models, the proposed method produces a large amount of real data, which may facilitate training results and avoid adverse effects engendered by artifactual data. We applied the constructed annotation dataset to train a deep learning you-only-look-once (YOLO) model to perform vehicle detection on street intersection surveillance videos. Experimental results demonstrated that the accurate detection performance was improved from a mean average precision (mAP) of 83.99 to 88.03.

Homomorphic Encryption is one of the most promising techniques to deal with privacy concerns, which is raised by remote deep learning paradigm, and maintain high classification accuracy. However, homomorphic encryption-based solutions are characterized by high overhead in terms of both computation and communication, which limits their adoption in pervasive health monitoring applications with constrained client-side devices. In this paper, we propose PReDIHERO, an improved privacy-preserving solution for remote deep learning inferences based on homomorphic encryption. The proposed solution applies a reversible obfuscation technique that successfully protects sensitive information, and enhances the client-side overhead compared to the conventional homomorphic encryption approach. The solution tackles three main heavyweight client-side tasks, namely, encryption and transmission of private data, refreshing encrypted data, and outsourcing computation of activation functions. The efficiency of the client-side is evaluated on a healthcare dataset and compared to a conventional homomorphic encryption approach. The evaluation results show that PReDIHERO requires increasingly less time and storage in comparison to conventional solutions when inferences are requested. At two hundreds inferences, the improvement ratio could reach more than 30 times in terms of computation overhead, and more than 8 times in terms of communication overhead. The same behavior is observed in sequential data and batch inferences, as we record an improvement ratio of more than 100 times in terms of computation overhead, and more than 20 times in terms of communication overhead.

To meet the application need of dynamic visualization VR display of 3D time-varying field, this paper designed an immersive visualization VR system of 3D time-varying field based on the Unity 3D framework. To reduce visual confusion caused by 3D time-varying field flow line drawing and improve the quality and efficiency of visualization rendering drawing, deep learning was used to extract features from the mesoscale vortex of the 3D time-varying field. Moreover, the 3D flow line dynamic visualization drawing was implemented through the Unity Visual Effect Graph particle system.

We tackle the problem where a server owns a trained Machine Learning (ML) model and a client/user has an unclassified query that he wishes to classify in secure and private fashion using the server’s model. During the process the server learns nothing, while the user learns only his final classification and nothing else. Since several ML classification algorithms, such as deep neural networks, support vector machines-SVM (and hyperplane decisions in general), Logistic Regression, Naïve Bayes, etc., can be expressed in terms of matrix operations, initially we propose novel secure matrix operations as our building blocks. On top of them we build our secure and private ML classification algorithms under strict security and privacy requirements. As our underlying cryptographic primitives are shown to be resilient to quantum computer attacks, our algorithms are also suitable for the post-quantum world. Our theoretical analysis and extensive experimental evaluations show that our secure matrix operations, hence our secure ML algorithms build on top of them as well, outperform the state of the art schemes in terms of computation and communication costs. This makes our algorithms suitable for devices with limited resources that are often found in Industrial IoT (Internet of Things)

Fake news, frequently making use of tampered photos, has currently emerged as a global epidemic, mainly due to the widespread use of social media as a present alternative to traditional news outlets. This development is often due to the swiftly declining price of advanced cameras and phones, which prompts the simple making of computerized pictures. The accessibility and usability of picture-altering softwares make picture-altering or controlling processes significantly simple, regardless of whether it is for the blameless or malicious plan. Various investigations have been utilized around to distinguish this sort of controlled media to deal with this issue. This paper proposes an efficient technique of copy-move forgery detection using the deep learning method. Two deep learning models such as Buster Net and VGG with FPN are used here to detect copy move forgery in digital images. The two models' performance is evaluated using the CoMoFoD dataset. The experimental result shows that VGG with FPN outperforms the Buster Net model for detecting forgery in images with an accuracy of 99.8% whereas the accuracy for the Buster Net model is 96.9%.