Biblio

Internet of Things is nowadays growing faster than ever before. Operators are planning or already creating dedicated networks for this type of devices. There is a need to create dedicated solutions for this type of network, especially solutions related to information security. In this article we present a mechanism of security-aware routing, which takes into account the evaluation of trust in devices and packet flows. We use trust relationships between flows and network nodes to create secure SDN paths, not ignoring also QoS and energy criteria. The system uses SDN infrastructure, enriched with Cognitive Packet Networks (CPN) mechanisms. Routing decisions are made by Random Neural Networks, trained with data fetched with Cognitive Packets. The proposed network architecture, implementing the security-by-design concept, was designed and is being implemented within the SerIoT project to demonstrate secure networks for the Internet of Things (IoT).

The recent success of brain-inspired deep neural networks (DNNs) in solving complex, high-level visual tasks has led to rising expectations for their potential to match the human visual system. However, DNNs exhibit idiosyncrasies that suggest their visual representation and processing might be substantially different from human vision. One limitation of DNNs is that they are vulnerable to adversarial examples, input images on which subtle, carefully designed noises are added to fool a machine classifier. The robustness of the human visual system against adversarial examples is potentially of great importance as it could uncover a key mechanistic feature that machine vision is yet to incorporate. In this study, we compare the visual representations of white- and black-box adversarial examples in DNNs and humans by leveraging functional magnetic resonance imaging (fMRI). We find a small but significant difference in representation patterns for different (i.e. white- versus black-box) types of adversarial examples for both humans and DNNs. However, human performance on categorical judgment is not degraded by noise regardless of the type unlike DNN. These results suggest that adversarial examples may be differentially represented in the human visual system, but unable to affect the perceptual experience.

Current testing for Deep Neural Networks (DNNs) focuses on quantity of test cases but ignores diversity. To the best of our knowledge, DeepXplore is the first white-box framework for Deep Learning testing by triggering differential behaviors between multiple DNNs and increasing neuron coverage to improve diversity. Since it is based on multiple DNNs facing problems that (1) the framework is not friendly to a single DNN, (2) if incorrect predictions made by all DNNs simultaneously, DeepXplore cannot generate test cases. This paper presents Test4Deep, a white-box testing framework based on a single DNN. Test4Deep avoids mistakes of multiple DNNs by inducing inconsistencies between predicted labels of original inputs and that of generated test inputs. Meanwhile, Test4Deep improves neuron coverage to capture more diversity by attempting to activate more inactivated neurons. The proposed method was evaluated on three popular datasets with nine DNNs. Compared to DeepXplore, Test4Deep produced average 4.59% (maximum 10.49%) more test cases that all found errors and faults of DNNs. These test cases got 19.57% more diversity increment and 25.88% increment of neuron coverage. Test4Deep can further be used to improve the accuracy of DNNs by average up to 5.72% (maximum 7.0%).

Deep neural networks are susceptible to various inference attacks as they remember information about their training data. We design white-box inference attacks to perform a comprehensive privacy analysis of deep learning models. We measure the privacy leakage through parameters of fully trained models as well as the parameter updates of models during training. We design inference algorithms for both centralized and federated learning, with respect to passive and active inference attackers, and assuming different adversary prior knowledge. We evaluate our novel white-box membership inference attacks against deep learning algorithms to trace their training data records. We show that a straightforward extension of the known black-box attacks to the white-box setting (through analyzing the outputs of activation functions) is ineffective. We therefore design new algorithms tailored to the white-box setting by exploiting the privacy vulnerabilities of the stochastic gradient descent algorithm, which is the algorithm used to train deep neural networks. We investigate the reasons why deep learning models may leak information about their training data. We then show that even well-generalized models are significantly susceptible to white-box membership inference attacks, by analyzing state-of-the-art pre-trained and publicly available models for the CIFAR dataset. We also show how adversarial participants, in the federated learning setting, can successfully run active membership inference attacks against other participants, even when the global model achieves high prediction accuracies.

Wide adoption of artificial neural networks in various domains has led to an increasing interest in defending adversarial attacks against them. Preprocessing defense methods such as pixel discretization are particularly attractive in practice due to their simplicity, low computational overhead, and applicability to various systems. It is observed that such methods work well on simple datasets like MNIST, but break on more complicated ones like ImageNet under recently proposed strong white-box attacks. To understand the conditions for success and potentials for improvement, we study the pixel discretization defense method, including more sophisticated variants that take into account the properties of the dataset being discretized. Our results again show poor resistance against the strong attacks. We analyze our results in a theoretical framework and offer strong evidence that pixel discretization is unlikely to work on all but the simplest of the datasets. Furthermore, our arguments present insights why some other preprocessing defenses may be insecure.

In this paper, we consider one of the approaches to the study of the characteristics of an information system that is under the influence of various factors, and their management using neural networks and wavelet transforms based on determining the relationship between the modified state of the information system and the possibility of dynamic analysis of effects. At the same time, the process of influencing the information system includes the following components: impact on the components providing the functions of the information system; determination of the result of exposure; analysis of the result of exposure; response to the result of exposure. As an input signal, the characteristics of the means that affect are taken. The system includes an adaptive response unit, the input of which receives signals about the prerequisites for changes, and at the output, this unit generates signals for the inclusion of appropriate means to eliminate or compensate for these prerequisites or directly the changes in the information system.

As the next generation of the power system, smart grid develops towards automated and intellectualized. Along with the benefits brought by smart grids, e.g., improved energy conversion rate, power utilization rate, and power supply quality, are the security challenges. One of the most important issues in smart grids is to ensure reliable communication between the secondary equipment. The state-of-art method to ensure smart grid security is to detect cyber attacks by deep learning. However, due to the small number of negative samples, the performance of the detection system is limited. In this paper, we propose a novel approach that utilizes the Generative Adversarial Network (GAN) to generate abundant negative samples, which helps to improve the performance of the state-of-art detection system. The evaluation results demonstrate that the proposed method can effectively improve the performance of the detection system by 4%.

The paper describes the process of coding information in the heterogenic multi-core system on chip for virtual-based security designed For image processing, signal processing and neural networks emulation. The coding of information carried out in assembly language according to the GOST. This is an implementation of the GOST - a standard symmetric key block cipher has a 64-bit block size and 256-bit key size.

Video Steganography is an extension of image steganography where any kind of file in any extension is hidden into a digital video. The video content is dynamic in nature and this makes the detection of hidden data difficult than other steganographic techniques. The main motive of using video steganography is that the videos can store large amount of data in it. This paper focuses on security using the combination of hybrid neural networks and hash function for determining the best bits in the cover video to embed the secret data. For the embedding process, the cover video and the data to be hidden is uploaded. Then the hash algorithm and neural networks are applied to form the stego video. For the extraction process, the reverse process is applied and the secret data is obtained. All experiments are done using MatLab2016a software.

The potential risk of agricultural product supply chain is huge because of the complex attributes specific to it. Actually the safety incidents of edible agricultural product emerge frequently in recent years, which expose the fragility of the agricultural product supply chain. In this paper the possible risk factors in agricultural product supply chain is analyzed in detail, the agricultural product supply chain risk evaluation index system and evaluation model are established, and an empirical analysis is made using BP neural network method. The results show that the risk ranking of the simulated evaluation is consistent with the target value ranking, and the risk assessment model has a good generalization and extension ability, and the model has a good reference value for preventing agricultural product supply chain risk.

A novel supervised machine learning system is developed to classify network traffic whether it is malicious or benign. To find the best model considering detection success rate, combination of supervised learning algorithm and feature selection method have been used. Through this study, it is found that Artificial Neural Network (ANN) based machine learning with wrapper feature selection outperform support vector machine (SVM) technique while classifying network traffic. To evaluate the performance, NSL-KDD dataset is used to classify network traffic using SVM and ANN supervised machine learning techniques. Comparative study shows that the proposed model is efficient than other existing models with respect to intrusion detection success rate.

In this study, a systematic mapping study was conducted to systematically evaluate publications on Intrusion Detection Systems with Deep Learning. 6088 papers have been examined by using systematic mapping method to evaluate the publications related to this paper, which have been used increasingly in the Intrusion Detection Systems. The goal of our study is to determine which deep learning algorithms were used mostly in the algortihms, which criteria were taken into account for selecting the preferred deep learning algorithm, and the most searched topics of intrusion detection with deep learning algorithm model. Scientific studies published in the last 10 years have been studied in the IEEE Explorer, ACM Digital Library, Science Direct, Scopus and Wiley databases.

From signal processing to emerging deep neural networks, a range of applications exhibit intrinsic error resilience. For such applications, approximate computing opens up new possibilities for energy-efficient computing by producing slightly inaccurate results using greatly simplified hardware. Adopting this approach, a variety of basic arithmetic units, such as adders and multipliers, have been effectively redesigned to generate approximate results for many error-resilient applications.In this work, we propose SECO, an approximate exponential function unit (EFU). Exponentiation is a key operation in many signal processing applications and more importantly in spiking neuron models, but its energy-efficient implementation has been inadequately explored. We also introduce a cross-layer design method for SECO to optimize the energy-accuracy trade-off. At the algorithm level, SECO offers runtime scaling between energy efficiency and accuracy based on approximate Taylor expansion, where the error is minimized by optimizing parameters using discrete gradient descent at design time. At the circuit level, our error analysis method efficiently explores the design space to select the energy-accuracy-optimal approximate multiplier at design time. In tandem, the cross-layer design and runtime optimization method are able to generate energy-efficient and accurate approximate EFU designs that are up to 99.7% accurate at a power consumption of 3.73 pJ per exponential operation. SECO is also evaluated on the adaptive exponential integrate-and-fire neuron model, yielding only 0.002% timing error and 0.067% value error compared to the precise neuron model.

The article is devoted to the improvement of neural network means of recognition of emotions on human geometry, which are defined for use in information systems of general purpose. It is shown that modern means of emotional recognition are based on the usual networks of critical disadvantage, because there is a lack of accuracy of recognition under the influence of purchased, characteristic of general-purpose information systems. It is determined that the above remarks relate to the turning of the face and the size of the image. A typical approach to overcoming this disadvantage through training is unacceptable for all protection options that are inappropriate for reasons of duration and compilation of the required training sample. It is proposed to increase the accuracy of recognition by submitting an expert data model to the neural network. An appropriate method for representing expert knowledge is developed. A feature of the method is the use of productive rules and the PNN neural network. Experimental verification of the developed solutions has been carried out. The obtained results allow to increase the efficiency of the termination and disclosure of the set of age networks, the characteristics of which are not presented in the registered statistical data.

We investigate a method of building a closed domain intelligent conversational agent using deep neural networks. A conversational agent is a dialog system intended to converse with a human, with a coherent structure. Our conversational agent uses a retrieval based model that identifies the intent of the input user query and maps it to a knowledge base to return appropriate results. Human conversations are based on context, but existing conversational agents are context insensitive. To overcome this limitation, our system uses a simple stack based context identification and storage system. The conversational agent generates responses according to the current context of conversation. allowing more human-like conversations.

Computational Intelligence (CI) has a great potential in Security & Defense (S&D) applications. Nevertheless, such potential seems to be still under exploited. In this work we first review CI applications in the maritime domain, done in the past decades by NATO Nations. Then we discuss challenges and opportunities for CI in S&D. Finally we argue that a review of the academic training of military officers is highly recommendable, in order to allow them to understand, model and solve new problems, using CI techniques.

This paper investigates the use of deep reinforcement learning (DRL) in the design of a "universal" MAC protocol referred to as Deep-reinforcement Learning Multiple Access (DLMA). The design framework is partially inspired by the vision of DARPA SC2, a 3-year competition whereby competitors are to come up with a clean-slate design that "best share spectrum with any network(s), in any environment, without prior knowledge, leveraging on machine-learning technique". While the scope of DARPA SC2 is broad and involves the redesign of PHY, MAC, and Network layers, this paper's focus is narrower and only involves the MAC design. In particular, we consider the problem of sharing time slots among a multiple of time-slotted networks that adopt different MAC protocols. One of the MAC protocols is DLMA. The other two are TDMA and ALOHA. The DRL agents of DLMA do not know that the other two MAC protocols are TDMA and ALOHA. Yet, by a series of observations of the environment, its own actions, and the rewards - in accordance with the DRL algorithmic framework - a DRL agent can learn the optimal MAC strategy for harmonious co-existence with TDMA and ALOHA nodes. In particular, the use of neural networks in DRL (as opposed to traditional reinforcement learning) allows for fast convergence to optimal solutions and robustness against perturbation in hyper- parameter settings, two essential properties for practical deployment of DLMA in real wireless networks.

Recent insights on language and vision with neural networks have been successfully applied to simple single-image visual question answering. However, to tackle real-life question answering problems on multimedia collections such as personal photos, we have to look at whole collections with sequences of photos or videos. When answering questions from a large collection, a natural problem is to identify snippets to support the answer. In this paper, we describe a novel neural network called Focal Visual-Text Attention network (FVTA) for collective reasoning in visual question answering, where both visual and text sequence information such as images and text metadata are presented. FVTA introduces an end-to-end approach that makes use of a hierarchical process to dynamically determine what media and what time to focus on in the sequential data to answer the question. FVTA can not only answer the questions well but also provides the justifications which the system results are based upon to get the answers. FVTA achieves state-of-the-art performance on the MemexQA dataset and competitive results on the MovieQA dataset.

False alarm and miss are two general kinds of alarm errors and they can decrease operator's trust in the alarm system. Specifically, there are two different forms of trust in such systems, represented by two kinds of responses to alarms in this research. One is compliance and the other is reliance. Besides false alarm and miss, the two responses are differentially affected by properties of the alarm system, situational factors or operator factors. However, most of the existing studies have qualitatively analyzed the relationship between a single variable and the two responses. In this research, all available experimental studies are identified through database searches using keyword "compliance and reliance" without restriction on year of publication to December 2017. Six relevant studies and fifty-two sets of key data are obtained as the data base of this research. Furthermore, neural network is adopted as a tool to establish the quantitative relationship between multiple factors and the two forms of trust, respectively. The result will be of great significance to further study the influence of human decision making on the overall fault detection rate and the false alarm rate of the human machine system.

Early detection of new kinds of malware always plays an important role in defending the network systems. Especially, if intelligent protection systems could themselves detect an existence of new malware types in their system, even with a very small number of malware samples, it must be a huge benefit for the organization as well as the social since it help preventing the spreading of that kind of malware. To deal with learning from few samples, term ``one-shot learning'' or ``fewshot learning'' was introduced, and mostly used in computer vision to recognize images, handwriting, etc. An approach introduced in this paper takes advantage of One-shot learning algorithms in solving the malware classification problem by using Memory Augmented Neural Network in combination with malware's API calls sequence, which is a very valuable source of information for identifying malware behavior. In addition, it also use some advantages of the development in Natural Language Processing field such as word2vec, etc. to convert those API sequences to numeric vectors before feeding to the one-shot learning network. The results confirm very good accuracies compared to the other traditional methods.

The increasing amount of malware variants seen in the wild is causing problems for Antivirus Software vendors, unable to keep up by creating signatures for each. The methods used to develop a signature, static and dynamic analysis, have various limitations. Machine learning has been used by Antivirus vendors to detect malware based on the information gathered from the analysis process. However, adversarial examples can cause machine learning algorithms to miss-classify new data. In this paper we describe a method for malware analysis by converting malware binaries to images and then preparing those images for training within a Generative Adversarial Network. These unsupervised deep neural networks are not susceptible to adversarial examples. The conversion to images from malware binaries should be faster than using dynamic analysis and it would still be possible to link malware families together. Using the Generative Adversarial Network, malware detection could be much more effective and reliable.

In this paper we present a new approach, named DLGraph, for malware detection using deep learning and graph embedding. DLGraph employs two stacked denoising autoencoders (SDAs) for representation learning, taking into consideration computer programs' function-call graphs and Windows application programming interface (API) calls. Given a program, we first use a graph embedding technique that maps the program's function-call graph to a vector in a low-dimensional feature space. One SDA in our deep learning model is used to learn a latent representation of the embedded vector of the function-call graph. The other SDA in our model is used to learn a latent representation of the given program's Windows API calls. The two learned latent representations are then merged to form a combined feature vector. Finally, we use softmax regression to classify the combined feature vector for predicting whether the given program is malware or not. Experimental results based on different datasets demonstrate the effectiveness of the proposed approach and its superiority over a related method.

Modern industrial control systems (ICS) act as victims of cyber attacks more often in last years. These attacks are hard to detect and their consequences can be catastrophic. Cyber attacks can cause anomalies in the work of the ICS and its technological equipment. The presence of mutual interference and noises in this equipment significantly complicates anomaly detection. Moreover, the traditional means of protection, which used in corporate solutions, require updating with each change in the structure of the industrial process. An approach based on the machine learning for anomaly detection was used to overcome these problems. It complements traditional methods and allows one to detect signal correlations and use them for anomaly detection. Additional Tennessee Eastman Process Simulation Data for Anomaly Detection Evaluation dataset was analyzed as example of industrial process. In the course of the research, correlations between the signals of the sensors were detected and preliminary data processing was carried out. Algorithms from the most common techniques of machine learning (decision trees, linear algorithms, support vector machines) and deep learning models (neural networks) were investigated for industrial process anomaly detection task. It's shown that linear algorithms are least demanding on computational resources, but they don't achieve an acceptable result and allow a significant number of errors. Decision tree-based algorithms provided an acceptable accuracy, but the amount of RAM, required for their operations, relates polynomially with the training sample volume. The deep neural networks provided the greatest accuracy, but they require considerable computing power for internal calculations.

Hardware Trojans (HTs) are malicious modifications of the original circuits intended to leak information or cause malfunction. Based on the Side Channel Analysis (SCA) technology, a set of hardware Trojan detection platform is designed for RTL circuits on the basis of HSPICE power consumption simulation. Principal Component Analysis (PCA) algorithm is used to reduce the dimension of power consumption data. An intelligent neural networks (NN) algorithm based on Particle Swarm Optimization (PSO) is introduced to achieve HTs recognition. Experimental results show that the detection accuracy of PSO NN method is much better than traditional BP NN method.

Due to the recent technological development, home appliances and electric devices are equipped with high-performance hardware device. Since demand of hardware devices is increased, production base become internationalized to mass-produce hardware devices with low cost and hardware vendors outsource their products to third-party vendors. Accordingly, malicious third-party vendors can easily insert malfunctions (also known as "hardware Trojans'') into their products. In this paper, we design six kinds of hardware Trojans at a gate-level netlist, and apply a neural-network (NN) based hardware-Trojan detection method to them. The designed hardware Trojans are different in trigger circuits. In addition, we insert them to normal circuits, and detect hardware Trojans using a machine-learning-based hardware-Trojan detection method with neural networks. In our experiment, we learned Trojan-infected benchmarks using NN, and performed cross validation to evaluate the learned NN. The experimental results demonstrate that the average TPR (True Positive Rate) becomes 72.9%, the average TNR (True Negative Rate) becomes 90.0%.