Biblio

Recent advances in Deep Learning (DL) have enabled a paradigm shift to include machine intelligence in a wide range of autonomous tasks. As a result, a largely unexplored surface has opened up for attacks jeopardizing the integrity of DL models and hindering the success of autonomous systems. To enable ubiquitous deployment of DL approaches across various intelligent applications, we propose to develop architectural support for hardware implementation of secure and robust DL. Towards this goal, we leverage hardware/software co-design to develop a DL execution engine that supports algorithms specifically designed to defend against various attacks. The proposed framework is enhanced with two real-time defense mechanisms, securing both DL training and execution stages. In particular, we enable model-level Trojan detection to mitigate backdoor attacks and malicious behaviors induced on the DL model during training. We further realize real-time adversarial attack detection to avert malicious behavior during execution. The proposed execution engine is equipped with hardware-level IP protection and usage control mechanism to attest the legitimacy of the DL model mapped to the device. Our design is modular and can be tuned to task-specific demands, e.g., power, throughput, and memory bandwidth, by means of a customized hardware compiler. We further provide an accompanying API to reduce the nonrecurring engineering cost and ensure automated adaptation to various domains and applications.

With the development of globalization, more and more enterprises are involved in the supply chain network with increasingly complex structure. In this paper, enterprises and relations in the logistics network are abstracted as nodes and edges of the complex network. A graph model for a supply chain network to specified industry is constructed, and the Neo4j graph database is employed to store the graph data. This paper uses the theoretical research tool of complex network to model and analyze the supply chain, and designs a supply chain network evaluation system which include static and dynamic measurement indexes according to the statistical characteristics of complex network. In this paper both the static and dynamic resilience characteristics of the the constructed supply chain network are evaluated from the perspective of complex network. The numeric experimental simulations are conducted for validation. This research has practical and theoretical significance for enterprises to make strategies to improve the anti-risk capability of supply chain network based on logistics network information.

Robot networks are susceptible to fail under the presence of malicious or defective robots. Resilient networks in the literature require high connectivity and large communication ranges, leading to high energy consumption in the communication network. This paper presents robot formations with guaranteed resiliency that use smaller communication ranges than previous results in the literature. The formations can be built on triangular and square lattices in the plane, and cubic lattices in the three-dimensional space. We support our theoretical framework with simulations.

Visual reasoning between visual image and natural language description is a long-standing challenge in computer vision. While recent approaches offer a great promise by compositionality or relational computing, most of them are oppressed by the challenge of training with datasets containing only a limited number of images with ground-truth texts. Besides, it is extremely time-consuming and difficult to build a larger dataset by annotating millions of images with text descriptions that may very likely lead to a biased model. Inspired by the majority success of webly supervised learning, we utilize readily-available web images with its noisy annotations for learning a robust representation. Our key idea is to presume on web images and corresponding tags along with fully annotated datasets in learning with knowledge embedding. We present a two-stage approach for the task that can augment knowledge through an effective embedding model with weakly supervised web data. This approach learns not only knowledge-based embeddings derived from key-value memory networks to make joint and full use of textual and visual information but also exploits the knowledge to improve the performance with knowledge-based representation learning for applying other general reasoning tasks. Experimental results on two benchmarks show that the proposed approach significantly improves performance compared with the state-of-the-art methods and guarantees the robustness of our model against visual reasoning tasks and other reasoning tasks.

The significant development of Internet of Things (IoT) paradigm for monitoring the real-time applications using the wireless communication technologies leads to various challenges. The secure data transmission and privacy is one of the key challenges of IoT enabled Wireless Sensor Networks (WSNs) communications. Due to heterogeneity of attackers like Man-in-Middle Attack (MIMA), the present single layered security solutions are not sufficient. In this paper, the robust cross-layer trust computation algorithm for MIMA attacker detection proposed for IoT enabled WSNs called IoT enabled Cross-Layer Man-in-Middle Attack Detection System (IC-MADS). In IC-MADS, first robust clustering method proposed to form the clusters and cluster head (CH) preference. After clustering, for every sensor node, its trust value computed using the parameters of three layers such as MAC, Physical, and Network layers to protect the network communications in presence of security threats. The simulation results prove that IC-MADS achieves better protection against MIMA attacks with minimum overhead and energy consumption.

Adaptation of e-commerce in third world countries requires more secure computing facilities. Online data is vulnerable and susceptible to active attacks. Hundreds of security mechanisms and services have been proposed to curb this challenge. However, available security mechanisms, sufficiently strong, are heavy for the machines used. To secure online data where machines' processing power and memory are deficient, a Hybrid Encryption Standard (HES) is proposed. The HES is built on the Data Encryption Standard (DES) algorithm and its siblings. The component units of the DES are redesigned towards reduced demands for processing power and memory. Precisely, white box designs of IP tables, PC tables, Expansion tables, Rotation tables, S-boxes and P-boxes are proposed, all aimed at reducing the processing time and memory demands. Evaluation of the performance of the HES algorithm against the performance of the traditional DES algorithm reveal that the HES out-performs the DES with regards to speed, memory demands, and general acceptance by novice practitioners in the cryptography field. In addition, reproducibility and flexibility are attractive features of the HES over the DES.

Attribution methods can provide powerful insights into the reasons for a classifier's decision. We argue that a key desideratum of an explanation method is its robustness to input hyperparameters which are often randomly set or empirically tuned. High sensitivity to arbitrary hyperparameter choices does not only impede reproducibility but also questions the correctness of an explanation and impairs the trust of end-users. In this paper, we provide a thorough empirical study on the sensitivity of existing attribution methods. We found an alarming trend that many methods are highly sensitive to changes in their common hyperparameters e.g. even changing a random seed can yield a different explanation! Interestingly, such sensitivity is not reflected in the average explanation accuracy scores over the dataset as commonly reported in the literature. In addition, explanations generated for robust classifiers (i.e. which are trained to be invariant to pixel-wise perturbations) are surprisingly more robust than those generated for regular classifiers.

Mining of data is used to analyze facts to discover formerly unknown patterns, classifying and grouping the records. There are several crucial scalable statistics mining platforms that have been developed in latest years. RapidMiner is a famous open source software which can be used for advanced analytics, Weka and Orange are important tools of machine learning for classifying patterns with techniques of clustering and regression, whilst Knime is often used for facts preprocessing like information extraction, transformation and loading. This article encapsulates the most important and robust platforms.

The proliferation of artificial intelligence based systems in all walks of life raises concerns about their safety and robustness, especially for cyber-physical systems including multiple machine learning components. In this paper, we introduce robustness contracts as a framework for compositional specification and reasoning about the robustness of cyber-physical systems based on neural network (NN) components. Robustness contracts can encompass and generalize a variety of notions of robustness which were previously proposed in the literature. They can seamlessly apply to NN-based perception as well as deep reinforcement learning (RL)-enabled control applications. We present a sound and complete algorithm that can efficiently verify the satisfaction of a class of robustness contracts on NNs by leveraging notions from Lagrangian duality to identify system configurations that violate the contracts. We illustrate the effectiveness of our approach on the verification of NN-based perception systems and deep RL-based control systems.

Spiking Neural Networks (SNNs) claim to present many advantages in terms of biological plausibility and energy efficiency compared to standard Deep Neural Networks (DNNs). Recent works have shown that DNNs are vulnerable to adversarial attacks, i.e., small perturbations added to the input data can lead to targeted or random misclassifications. In this paper, we aim at investigating the key research question: "Are SNNs secure?" Towards this, we perform a comparative study of the security vulnerabilities in SNNs and DNNs w.r.t. the adversarial noise. Afterwards, we propose a novel black-box attack methodology, i.e., without the knowledge of the internal structure of the SNN, which employs a greedy heuristic to automatically generate imperceptible and robust adversarial examples (i.e., attack images) for the given SNN. We perform an in-depth evaluation for a Spiking Deep Belief Network (SDBN) and a DNN having the same number of layers and neurons (to obtain a fair comparison), in order to study the efficiency of our methodology and to understand the differences between SNNs and DNNs w.r.t. the adversarial examples. Our work opens new avenues of research towards the robustness of the SNNs, considering their similarities to the human brain's functionality.

In recent years, ICN (Information-Centric Networking) has been under the spotlight as a network that mainly focuses on transmitted and received data rather than on the hosts that transmit and receive data. Generally, the communication networks such as ICNs are required to be robust against network failures caused by attacks and disasters. One of the metrics for the robustness of conventional host-centric networks, e.g., TCP/IP network, is reachability between nodes in the network after network failures, whereas the key metric for the robustness of ICNs is content availability. In this paper, we focus on an arbitrary ICN network and derive the content availability for a given probability of node removal. Especially, we analytically obtain the average content availability over an entire network in the case where just a single path from a node to a repository, i.e., contents server, storing contents is available and where multiple paths to the repository are available, respectively. Furthermore, through several numerical evaluations, we investigate the effect of the structure of network topology as well as the pattern and scale of the network failures on the content availability in ICN. Our findings include that, regardless of patterns of network failures, the content availability is significantly improved by caching contents at routers and using multiple paths, and that the content availability is more degraded at cluster-based node removal compared with random node removal.

For the purpose of outdoor advertising market researching, AD images are recaptured and uploaded everyday for statistics. But the quality of the recaptured advertising images are often affected by conditions such as angle, distance, and light during the shooting process, which consequently reduce either the speed or the accuracy of the retrieving algorithm. In this paper, we proposed a hash retrieval method based on convolutional neural networks for recaptured images. The basic idea is to add a hash layer to the convolutional neural network and then extract the binary hash code output by the hash layer to perform image retrieval in lowdimensional Hamming space. Experimental results show that the retrieval performance is improved compared with the current commonly used hash retrieval methods.

Clustering is a concept in data mining, which divides a data set into different classes or clusters according to a specific standard, making the similarity of data objects in the same cluster as large as possible. Clustering by fast search and find of density peaks (DPC) is a novel clustering algorithm based on density. It is simple and novel, only requiring fewer parameters to achieve better clustering effect, without the requirement for iterative solution. And it has expandability and can detect the clustering of any shape. However, DPC algorithm still has some defects, such as it employs the clear neighborhood relations to calculate local density, so it cannot identify the neighborhood membership of different values of points from the distance of points and It is impossible to accurately cluster the data of the multi-density peak. The fuzzy neighborhood density peak clustering algorithm is proposed for this shortcoming (F-DPC): novel local density is defined by the fuzzy neighborhood relationship. The fuzzy set theory can be used to make the fuzzy neighborhood function of local density more sensitive, so that the clustering for data set of various shapes and densities is more robust. Experiments show that the algorithm has high accuracy and robustness.

The increased reliance on the Internet and the corresponding surge in connectivity demand has led to a significant growth in Internet-of-Things (IoT) devices. The continued deployment of IoT devices has in turn led to an increase in network attacks due to the larger number of potential attack surfaces as illustrated by the recent reports that IoT malware attacks increased by 215.7% from 10.3 million in 2017 to 32.7 million in 2018. This illustrates the increased vulnerability and susceptibility of IoT devices and networks. Therefore, there is a need for proper effective and efficient attack detection and mitigation techniques in such environments. Machine learning (ML) has emerged as one potential solution due to the abundance of data generated and available for IoT devices and networks. Hence, they have significant potential to be adopted for intrusion detection for IoT environments. To that end, this paper proposes an optimized ML-based framework consisting of a combination of Bayesian optimization Gaussian Process (BO-GP) algorithm and decision tree (DT) classification model to detect attacks on IoT devices in an effective and efficient manner. The performance of the proposed framework is evaluated using the Bot-IoT-2018 dataset. Experimental results show that the proposed optimized framework has a high detection accuracy, precision, recall, and F-score, highlighting its effectiveness and robustness for the detection of botnet attacks in IoT environments.

Adversarial examples are known as carefully perturbed images fooling image classifiers. We propose a geometric framework to generate adversarial examples in one of the most challenging black-box settings where the adversary can only generate a small number of queries, each of them returning the top-1 label of the classifier. Our framework is based on the observation that the decision boundary of deep networks usually has a small mean curvature in the vicinity of data samples. We propose an effective iterative algorithm to generate query-efficient black-box perturbations with small p norms which is confirmed via experimental evaluations on state-of-the-art natural image classifiers. Moreover, for p=2, we theoretically show that our algorithm actually converges to the minimal perturbation when the curvature of the decision boundary is bounded. We also obtain the optimal distribution of the queries over the iterations of the algorithm. Finally, experimental results confirm that our principled black-box attack algorithm performs better than state-of-the-art algorithms as it generates smaller perturbations with a reduced number of queries.

The increase of cyber attacks in both the numbers and varieties in recent years demands to build a more sophisticated network intrusion detection system (NIDS). These NIDS perform better when they can monitor all the traffic traversing through the network like when being deployed on a Software-Defined Network (SDN). Because of the inability to detect zero-day attacks, signature-based NIDS which were traditionally used for detecting malicious traffic are beginning to get replaced by anomaly-based NIDS built on neural networks. However, recently it has been shown that such NIDS have their own drawback namely being vulnerable to the adversarial example attack. Moreover, they were mostly evaluated on the old datasets which don't represent the variety of attacks network systems might face these days. In this paper, we present Reconstruction from Partial Observation (RePO) as a new mechanism to build an NIDS with the help of denoising autoencoders capable of detecting different types of network attacks in a low false alert setting with an enhanced robustness against adversarial example attack. Our evaluation conducted on a dataset with a variety of network attacks shows denoising autoencoders can improve detection of malicious traffic by up to 29% in a normal setting and by up to 45% in an adversarial setting compared to other recently proposed anomaly detectors.

It is now possible to synthesize highly realistic images of people who do not exist. Such content has, for example, been implicated in the creation of fraudulent socialmedia profiles responsible for dis-information campaigns. Significant efforts are, therefore, being deployed to detect synthetically-generated content. One popular forensic approach trains a neural network to distinguish real from synthetic content.We show that such forensic classifiers are vulnerable to a range of attacks that reduce the classifier to near- 0% accuracy. We develop five attack case studies on a state- of-the-art classifier that achieves an area under the ROC curve (AUC) of 0.95 on almost all existing image generators, when only trained on one generator. With full access to the classifier, we can flip the lowest bit of each pixel in an image to reduce the classifier's AUC to 0.0005; perturb 1% of the image area to reduce the classifier's AUC to 0.08; or add a single noise pattern in the synthesizer's latent space to reduce the classifier's AUC to 0.17. We also develop a black-box attack that, with no access to the target classifier, reduces the AUC to 0.22. These attacks reveal significant vulnerabilities of certain image-forensic classifiers.

Cybersecurity community is slowly leveraging Machine Learning (ML) to combat ever evolving threats. One of the biggest drivers for successful adoption of these models is how well domain experts and users are able to understand and trust their functionality. As these black-box models are being employed to make important predictions, the demand for transparency and explainability is increasing from the stakeholders.Explanations supporting the output of ML models are crucial in cyber security, where experts require far more information from the model than a simple binary output for their analysis. Recent approaches in the literature have focused on three different areas: (a) creating and improving explainability methods which help users better understand the internal workings of ML models and their outputs; (b) attacks on interpreters in white box setting; (c) defining the exact properties and metrics of the explanations generated by models. However, they have not covered, the security properties and threat models relevant to cybersecurity domain, and attacks on explainable models in black box settings.In this paper, we bridge this gap by proposing a taxonomy for Explainable Artificial Intelligence (XAI) methods, covering various security properties and threat models relevant to cyber security domain. We design a novel black box attack for analyzing the consistency, correctness and confidence security properties of gradient based XAI methods. We validate our proposed system on 3 security-relevant data-sets and models, and demonstrate that the method achieves attacker's goal of misleading both the classifier and explanation report and, only explainability method without affecting the classifier output. Our evaluation of the proposed approach shows promising results and can help in designing secure and robust XAI methods.

Resilience is regarded nowadays as the ideal solution that can be envisaged by sociotechnical systems for coping with potential threats and crises. This being said, gaining and maintaining this ability is not always easy, given the multitude of risks driving the adverse and challenging events. This paper aims to propose a method consecrated to the assessment of risks directly affecting resilience. This work is conducted within the framework of risk assessment and resilience engineering approaches. A 5×5 matrix, dedicated to the identification and assessment of risk factors that constitute threats to the system resilience, has been elaborated. This matrix consists of two axes, namely, the impact on resilience metrics and the availability and effectiveness of resilience planning. Checklists serving to collect information about these two attributes are established and a case study is undertaken. In this paper, a new method for identifying and assessing risk factors menacing directly the resilience of a given system is presented. The analysis of these risks must be given priority to make the system more resilient to shocks.

Proof of integrity in produced video data by surveillance cameras requires active forensic methods such as signatures, otherwise authenticity and integrity can be comprised and data becomes unusable e. g. for legal evidence. But a simple file- or stream-signature loses its validity when the stream is cut in parts or by separating data and signature. Using the principles of security in distributed systems similar to those of blockchain and distributed ledger technologies (BC/DLT), a chain which consists of the frames of a video which frame hash values will be distributed among a camera sensor network is presented. The backbone of this Framechain within the camera sensor network will be a camera identity concept to ensure accountability, integrity and authenticity according to the extended CIA triad security concept. Modularity by secure sequences, autarky in proof and robustness against natural modulation of data are the key parameters of this new approach. It allows the standalone data and even parts of it to be used as hard evidence.

Named Data Networking (NDN) is a content-centric networking, where the publisher of the packet signs and encapsulates the data packet with a name-content-signature encryption to verify the authenticity and integrity of itself. This scheme can solve many of the security issues inherently compared to IP networking. NDN also support mobility since it hides the point-to-point connection details. However, an extreme attack takes place when an NDN consumer newly connects to a network. A Man-in-the-middle (MITM) malicious node can block the consumer and keep intercepting the interest packets sent out so as to fake the corresponding data packets signed with its own private key. Without knowledge and trust to the network, the NDN consumer can by no means perceive the attack and thus exposed to severe security and privacy hazard. In this paper, the Name-Signature Lookup System (NSLS) and corresponding Name-Signature Lookup Protocol (NSLP) is introduced to verify packets with their registered genuine publisher even in an untrusted network with the help of embedded keys inside Network Interface Controller (NIC), by which attacks like MITM is eliminated. A theoretical analysis of comparing NSLS with existing security model is provided. Digest algorithm SHA-256 and signature algorithm RSA are used in the NSLP model without specific preference.

Network covert timing channel(NCTC) is a process of transmitting hidden information by means of inter-packet delay (IPD) of legitimate network traffic. Their ability to evade traditional security policies makes NCTCs a grave security concern. However, a robust method that can be used to detect a large number of NCTCs is missing. In this paper, a NCTC detection method based on chaos theory and threshold secret sharing is proposed. Our method uses chaos theory to reconstruct a high-dimensional phase space from one-dimensional time series and extract the unique and stable channel traits. Then, a channel identifier is constructed using the secret reconstruction strategy from threshold secret sharing to realize the mapping of the channel features to channel identifiers. Experimental results show that the approach can detect varieties of NCTCs with a guaranteed true positive rate and greatly improve the versatility and robustness.

Artificial Intelligence (AI) is influencing almost all areas of human life. Even though these AI-based systems frequently provide state-of-the-art performance, humans still hesitate to develop, deploy, and use AI systems. The main reason for this is the lack of trust in AI systems caused by the deficiency of transparency of existing AI systems. As a solution, “Trustworthy AI” research area merged with the goal of defining guidelines and frameworks for improving user trust in AI systems, allowing humans to use them without fear. While trust in AI is an active area of research, very little work exists where the focus is to build human trust to improve the interactions between human and AI systems. In this paper, we provide a concise survey on concepts of trustworthy AI. Further, we present trustworthy AI development guidelines for improving the user trust to enhance the interactions between AI systems and humans, that happen during the AI system life cycle.

Increasing deep neural networks are applied in solving graph evolved tasks, such as node classification and link prediction. However, the vulnerability of deep models can be revealed using carefully crafted adversarial examples generated by various adversarial attack methods. To explore this security problem, we define the link prediction adversarial attack problem and put forward a novel iterative gradient attack (IGA) strategy using the gradient information in the trained graph autoencoder (GAE) model. Not surprisingly, GAE can be fooled by an adversarial graph with a few links perturbed on the clean one. The results on comprehensive experiments of different real-world graphs indicate that most deep models and even the state-of-the-art link prediction algorithms cannot escape the adversarial attack, such as GAE. We can benefit the attack as an efficient privacy protection tool from the link prediction of unknown violations. On the other hand, the adversarial attack is a robust evaluation metric for current link prediction algorithms of their defensibility.

The Global Positioning System (GPS) can determine the position of any person or object on earth based on satellite signals. But when inside the building, the GPS cannot receive signals, the indoor positioning system will determine the precise position. How to achieve more precise positioning is the difficulty of an indoor positioning system now. In this paper, we proposed an ultra-wideband fingerprinting positioning method based on a convolutional neural network (CNN), and we collect the dataset in a room to test the model, then compare our method with the existing method. In the experiment, our method can reach an accuracy of 98.36%. Compared with other fingerprint positioning methods our method has a great improvement in robustness. That results show that our method has good practicality while achieves higher accuracy.