Biblio

Largely known for attack scenarios, code reuse techniques at a closer look reveal properties that are appealing also for program obfuscation. We explore the popular return-oriented programming paradigm under this light, transforming program functions into ROP chains that coexist seamlessly with the surrounding software stack. We show how to build chains that can withstand popular static and dynamic deobfuscation approaches, evaluating the robustness and overheads of the design over common programs. The results suggest a significant amount of computational resources would be required to carry a deobfuscation attack for secret finding and code coverage goals.

Deep Learning (DL) algorithms have gained popularity owing to their practical problem-solving capacity. However, they suffer from a serious integrity threat, i.e., their vulnerability to adversarial attacks. In the quest for DL trustworthiness, recent works claimed the inherent robustness of Spiking Neural Networks (SNNs) to these attacks, without considering the variability in their structural spiking parameters. This paper explores the security enhancement of SNNs through internal structural parameters. Specifically, we investigate the SNNs robustness to adversarial attacks with different values of the neuron's firing voltage thresholds and time window boundaries. We thoroughly study SNNs security under different adversarial attacks in the strong white-box setting, with different noise budgets and under variable spiking parameters. Our results show a significant impact of the structural parameters on the SNNs' security, and promising sweet spots can be reached to design trustworthy SNNs with 85% higher robustness than a traditional non-spiking DL system. To the best of our knowledge, this is the first work that investigates the impact of structural parameters on SNNs robustness to adversarial attacks. The proposed contributions and the experimental framework is available online 11https://github.com/rda-ela/SNN-Adversarial-Attacks to the community for reproducible research.

Robustness, both to accident and to malevolent perturbations, is a crucial determinant of the successful deployment of deep convolutional neural network based SAR ATR systems in various security-sensitive applications. This paper performs a detailed adversarial robustness evaluation of deep convolutional neural network based SAR ATR models across two public available SAR target recognition datasets. For each model, seven different adversarial perturbations, ranging from gradient based optimization to self-supervised feature distortion, are generated for each testing image. Besides adversarial average recognition accuracy, feature attribution techniques have also been adopted to analyze the feature diffusion effect of adversarial attacks, which promotes the understanding of vulnerability of deep learning models.

Vehicular Ad Hoc Networks (VANETs) have the potential to improve road safety and reduce traffic congestion by enhancing sharing of messages about road conditions. Communication in VANETs depends upon a Public Key Infrastructure (PKI) that checks for message confidentiality, integrity, and authentication. One challenge that the PKI infrastructure does not eliminate is the possibility of malicious vehicles mounting a Distributed Denial of Service (DDoS) attack. We present a scheme that combines statistical modeling and machine learning techniques to detect and prevent blackhole attacks in a VANET environment.Simulation results demonstrate that on average, our model produces an Area Under The Curve (ROC) and Receiver Operating Characteristics (AUC) score of 96.78% which is much higher than a no skill ROC AUC score and only 3.22% away from an ideal ROC AUC score. Considering all the performance metrics, we show that the Support Vector Machine (SVM) and Gradient Boosting classifier are more accurate and perform consistently better under various circumstances. Both have an accuracy of over 98%, F1-scores of over 95%, and ROC AUC scores of over 97%. Our scheme is robust and accurate as evidenced by its ability to identify and prevent blackhole attacks. Moreover, the scheme is scalable in that addition of vehicles to the network does not compromise its accuracy and robustness.

Graph pooling is an essential ingredient of Graph Neural Networks (GNNs) in graph classification and regression tasks. For these tasks, different pooling strategies have been proposed to generate a graph-level representation by downsampling and summarizing nodes' features in a graph. However, most existing pooling methods are unable to capture distinguishable structural information effectively. Besides, they are prone to adversarial attacks. In this work, we propose a novel pooling method named as HIBPool where we leverage the Information Bottleneck (IB) principle that optimally balances the expressiveness and robustness of a model to learn representations of input data. Furthermore, we introduce a novel structure-aware Discriminative Pooling Readout (DiP-Readout) function to capture the informative local subgraph structures in the graph. Finally, our experimental results show that our model significantly outperforms other state-of-art methods on several graph classification benchmarks and more resilient to feature-perturbation attack than existing pooling methods11Source code at: https://github.com/forkkr/HIBPool.

Random K-out graphs, denoted \$$\backslash$mathbbH(n;K)\$, are generated by each of the \$n\$ nodes drawing \$K\$ out-edges towards \$K\$ distinct nodes selected uniformly at random, and then ignoring the orientation of the arcs. Recently, random K-out graphs have been used in applications as diverse as random (pairwise) key predistribution in ad-hoc networks, anonymous message routing in crypto-currency networks, and differentially-private federated averaging. In many applications, connectivity of the random K-out graph when some of its nodes are dishonest, have failed, or have been captured is of practical interest. We provide a comprehensive set of results on the connectivity and giant component size of \$$\backslash$mathbbH(n;K\_n,$\backslash$gamma\_n)\$, i.e., random K-out graph when \textsubscriptn of its nodes, selected uniformly at random, are deleted. First, we derive conditions for \textsubscriptn and \$n\$ that ensure, with high probability (whp), the connectivity of the remaining graph when the number of deleted nodes is \$$\backslash$gamma\_n=Ømega(n)\$ and \$$\backslash$gamma\_n=o(n)\$, respectively. Next, we derive conditions for \$$\backslash$mathbbH(n;K\_n, $\backslash$gamma\_n)\$ to have a giant component, i.e., a connected subgraph with \$Ømega(n)\$ nodes, whp. This is also done for different scalings of \textsubscriptn and upper bounds are provided for the number of nodes outside the giant component. Simulation results are presented to validate the usefulness of the results in the finite node regime.

The whole Software-Defined Networking (SDN) system might be out of service when the control plane is overloaded by control plane saturation attacks. In this attack, a malicious host can manipulate massive table-miss packets to exhaust the control plane resources. Even though many studies have focused on this problem, systems still suffer from more influenced switches because of centralized mitigation policies, and long recovery delay because of the remaining attack flows. To solve these problems, we propose FSDM, a Fast recovery Saturation attack Detection and Mitigation framework. For detection, FSDM extracts the distribution of Control Channel Occupation Rate (CCOR) to detect the attack and locates the port that attackers come from. For mitigation, with the attacker's location and distributed Mitigation Agents, FSDM adopts different policies to migrate or block attack flows, which influences fewer switches and protects the control plane from resource exhaustion. Besides, to reduce the system recovery delay, FSDM equips a novel functional module called Force\_Checking, which enables the whole system to quickly clean up the remaining attack flows and recovery faster. Finally, we conducted extensive experiments, which show that, with the increasing of attack PPS (Packets Per Second), FSDM only suffers a minor recovery delay increase. Compared with traditional methods without cleaning up remaining flows, FSDM saves more than 81% of ping RTT under attack rate ranged from 1000 to 4000 PPS, and successfully reduced the delay of 87% of HTTP requests time under large attack rate ranged from 5000 to 30000 PPS.

Although quite a few results on resilient consensus of multi-agent systems with malicious agents and fixed topology have been reported in the literature, we lack any known results on such a problem for multi-agent systems with time-varying topologies. Herein, we study the resilient consensus problem of time-varying networked systems in the presence of misbehaving nodes. A novel concept of joint ( r, s) -robustness is firstly proposed to characterize the robustness of the time-varying topologies. It is further revealed that the resilient consensus of multi-agent systems under F-total malicious network can be reached by the Weighted Mean-Subsequence-Reduced algorithm if and only if the time-varying graph is jointly ( F+1, F+1) -robust. Numerical simulations are finally performed to verify the effectiveness of the analytical results.

Despite the performance of state-of-the-art Artificial Intelligence (AI) systems, some sectors hesitate to adopt AI because of a lack of trust in these systems. This attitude is prevalent among high-risk areas, where there is a reluctance to remove humans entirely from the loop. In these scenarios, Augmentation provides a preferred alternative over complete Automation. Instead of replacing humans, AI Augmentation uses AI to improve and support human operations, creating an environment where humans work side by side with AI systems. In this paper, we discuss how AI Augmentation can provide a path for building Trustworthy AI. We exemplify this approach using Robot Teleoperation. We lay out design guidelines and motivations for the development of AI Augmentation for Robot Teleoperation. Finally, we discuss the design of a Robot Teleoperation testbed for the development of AI Augmentation systems.

Existing Lipschitz-based provable defences to adversarial examples only cover the L2 threat model. We introduce the first bound that makes use of Lipschitz continuity to provide a more general guarantee for threat models based on any Lp norm. Additionally, a new strategy is proposed for designing network architectures that exhibit superior provable adversarial robustness over conventional convolutional neural networks. Experiments are conducted to validate our theoretical contributions, show that the assumptions made during the design of our novel architecture hold in practice, and quantify the empirical robustness of several Lipschitz-based adversarial defence methods.

Sometimes a security-critical decision must be made using information provided by peers. Think of routing messages, user reports, sensor data, navigational information, blockchain updates. Attackers manifest as peers that strategically report fake information. Trust models use the provided information, and attempt to suggest the correct decision. A model that appears accurate by empirical evaluation of attacks may still be susceptible to manipulation. For a security-critical decision, it is important to take the entire attack space into account. Therefore, we define the property of robustness: the probability of deciding correctly, regardless of what information attackers provide. We introduce the notion of realisations of honesty, which allow us to bypass reasoning about specific feedback. We present two schemes that are optimally robust under the right assumptions. The “majority-rule” principle is a special case of the other scheme which is more general, named “most plausible realisations”.

DNN models have suffered from adversarial example attacks, which lead to inconsistent prediction results. As opposed to the gradient-based attack, which assumes white-box access to the model by the attacker, we focus on more realistic input perturbations from the real-world and their actual impact on the model robustness without any presence of the attackers. In this work, we promote a standardized framework to quantify the robustness against real-world threats. It is composed of a set of safety properties associated with common violations, a group of metrics to measure the minimal perturbation that causes the offense, and various criteria that reflect different aspects of the model robustness. By revealing comparison results through this framework among 13 pre-trained ImageNet classifiers, three state-of-the-art object detectors, and three cloud-based content moderators, we deliver the status quo of the real-world model robustness. Beyond that, we provide robustness benchmarking datasets for the community.

False data injection attack (FDIA) is a real threat to smart grids due to its wide range of vulnerabilities and impacts. Designing a proper detection scheme for FDIA is the 1stcritical step in defending the attack in smart grids. In this paper, we investigate two main statistical techniques-based approaches in this regard. The first is based on the principal component analysis (PCA), and the second is based on the canonical correlation analysis (CCA). The test cases illustrate a better characterization performance of FDIA using CCA compared to the PCA. Further, CCA provides a better differentiation of FDIA from normal grid contingencies. On the other hand, PCA provides a significantly reduced false alarm rate.

Radio frequency identification (RFID) is widespread and still necessary in many important applications. However, and in various significant cases, the use of this technology faces multiple security issues that must be addressed. This is mainly related to the use of RFID tags (transponders) which are electronic components communicating wirelessly, and hence they are vulnerable to multiple attacks through several means. In this work, an extensive fault analysis is performed on a tag architecture in order to evaluate its hardness. Tens of millions of single-bit upset (SBU) and multiple-bit upset (MBU) faults are emulated randomly on this tag architecture using an FPGA-based emulation platform. The emulated faults are classified under five groups according to faults effect on the tag behaviour. The obtained results show the faults effect variation in function of the number of MBU affected bits. The interpretation of this variation allows evaluating the tag robustness. The proposed approach represents an efficient mean that permits to study tag architectures at the design level and evaluating their robustness and vulnerability to fault attacks.

Authenticate on the system using only the authentication method based on username and password is not enough to ensure an acceptable level of information security for a critical system. It has been used in a multi factor authentication to increase the information security during the authentication process. However factors like what you have cause an inconvenience to the users, because the users during the authentication process always will need to have a device in their possession that complements the authentication process. By the other side of the biometric factor might change during the time, it needs an auxiliary device that will increase the costs and it also might be dependent from environmental conditions to work appropriately. To avoid some problems that exist in multi factor authentication, this work purposes authentication through semantic representation in OWL (web Ontology Language) tuples of recognized concepts in images as a form to increase the security in the authentication process. A proof of the concept was modeled and implemented, it has a demonstration that the robustness of this authentication system depends on the complexity of relationship in the semantic base (ontology) and in the simplicity of the relationship identified in the images.

Malware classification remains at the forefront of ongoing research as the prevalence of metamorphic malware introduces new challenges to anti-virus vendors and firms alike. One approach to malware classification is Static Analysis - a form of analysis which does not require malware to be executed before classification can be performed. For this reason, a lightweight classifier based on the features of a malware binary is preferred, with relatively low computational overhead. In this work a modified convolutional neural network (CNN) architecture was deployed which integrated a complexity-based evaluation based on box-counting. This was implemented by setting up max-pooling layers in parallel, and then extracting the fractal dimension using a polyscalar relationship based on the resolution of the measurement scale and the number of elements of a malware image covered in the measurement under consideration. To test the robustness and efficacy of our approach we trained and tested on over 9300 malware binaries from 25 unique malware families. This work was compared to other award-winning image recognition models, and results showed categorical accuracy in excess of 96.54%.

Recent advances in adversarial attacks uncover the intrinsic vulnerability of modern deep neural networks. Since then, extensive efforts have been devoted to enhancing the robustness of deep networks via specialized learning algorithms and loss functions. In this work, we take an architectural perspective and investigate the patterns of network architectures that are resilient to adversarial attacks. To obtain the large number of networks needed for this study, we adopt one-shot neural architecture search, training a large network for once and then finetuning the sub-networks sampled therefrom. The sampled architectures together with the accuracies they achieve provide a rich basis for our study. Our ''robust architecture Odyssey'' reveals several valuable observations: 1) densely connected patterns result in improved robustness; 2) under computational budget, adding convolution operations to direct connection edge is effective; 3) flow of solution procedure (FSP) matrix is a good indicator of network robustness. Based on these observations, we discover a family of robust architectures (RobNets). On various datasets, including CIFAR, SVHN, Tiny-ImageNet, and ImageNet, RobNets exhibit superior robustness performance to other widely used architectures. Notably, RobNets substantially improve the robust accuracy ( 5% absolute gains) under both white-box and black-box attacks, even with fewer parameter numbers. Code is available at https://github.com/gmh14/RobNets.

Cloud computing is applied in various domains, among which social recommender systems are well-received because of their effectivity to provide suggestions for users. Social recommender systems perform well in alleviating cold start problem, but it suffers from shilling attack due to its natural openness. Shilling attack is an injection attack mainly acting on the training process of machine learning, which aims to advance or suppress the recommendation ranking of target items. Some researchers have studied the influence of shilling attacks in two perspectives simultaneously, which are user-item's rating and user-user's relation. However, they take more consideration into user-item's rating, and up to now, the construction of user-user's relation has not been explored in depth. To explore shilling attacks with complex relations, in this paper, we propose two novel attack models based on triangle relations in social networks. Furthermore, we explore the influence of these models on five social recommendation algorithms. The experimental results on three datasets show that the recommendation can be affected by the triangle relation attacks. The attack model combined with triangle relation has a better attack effect than the model only based on rating injection and the model combined with random relation. Besides, we compare the functions of triangle relations in friend recommendation and product recommendation.

The advancing information technology is playing more and more important role in data mining of relational database.1 The transfer and sharing of databases cause the copyright-related security threats. Database watermarking technology can effectively solve the problem with copyright protection and traceability, which has been attracting researchers' attention. In this paper, we proposed a novel, robust and reversible database watermarking technique, named histogram shifting watermarking based on random forest and genetic algorithm (RF-GAHCSW). It greatly improves the watermark capacity by means of histogram width reduction and eliminates the impact of the prediction error attack. Meanwhile, random forest algorithm is used to select important attributes for watermark embedding, and genetic algorithm is employed to find the optimal secret key for the database grouping and determine the position of watermark embedding to improve the watermark capacity and reduce data distortion. The experimental results show that the robustness of RF-GAHCSW is greatly improved, compared with the original HSW, and the distortion has little effect on the usability of database.

Person re-identification (re-ID) is a key problem in smart supervision of camera networks. Over the past years, models using deep learning have become state of the art. However, it has been shown that deep neural networks are flawed with adversarial examples, i.e. human-imperceptible perturbations. Extensively studied for the task of image closed- set classification, this problem can also appear in the case of open-set retrieval tasks. Indeed, recent work has shown that we can also generate adversarial examples for metric learning systems such as re-ID ones. These models remain vulnerable: when faced with adversarial examples, they fail to correctly recognize a person, which represents a security breach. These attacks are all the more dangerous as they are impossible to detect for a human operator. Attacking a metric consists in altering the distances between the feature of an attacked image and those of reference images, i.e. guides. In this article, we investigate different possible attacks depending on the number and type of guides available. From this metric attack family, two particularly effective attacks stand out. The first one, called Self Metric Attack, is a strong attack that does not need any image apart from the attacked image. The second one, called FurthestNegative Attack, makes full use of a set of images. Attacks are evaluated on commonly used datasets: Market1501 and DukeMTMC. Finally, we propose an efficient extension of adversarial training protocol adapted to metric learning as a defense that increases the robustness of re-ID models.1

Deep learning plays an increasingly important role in various fields due to its superior performance, and it also achieves advanced recognition performance in the field of image classification. However, the vulnerability of deep learning in the adversarial environment cannot be ignored, and the prediction result of the model is likely to be affected by the small perturbations added to the samples by the adversary. In this paper, we propose a two-layer dynamic defense method based on defensive techniques pool and retrained branch model pool. First, we randomly select defense methods from the defense pool to process the input. The perturbation ability of the adversarial samples preprocessed by different defense methods changed, which would produce different classification results. In addition, we conduct adversarial training based on the original model and dynamically generate multiple branch models. The classification results of these branch models for the same adversarial sample is inconsistent. We can detect the adversarial samples by using the inconsistencies in the output results of the two layers. The experimental results show that the two-layer dynamic defense method we designed achieves a good defense effect.

In this paper, we study the resilient vector consensus problem in multi-agent networks and improve resilience guarantees of existing algorithms. In resilient vector consensus, agents update their states, which are vectors in ℝd, by locally interacting with other agents some of which might be adversarial. The main objective is to ensure that normal (non-adversarial) agents converge at a common state that lies in the convex hull of their initial states. Currently, resilient vector consensus algorithms, such as approximate distributed robust convergence (ADRC) are based on the idea that to update states in each time step, every normal node needs to compute a point that lies in the convex hull of its normal neighbors' states. To compute such a point, the idea of Tverberg partition is typically used, which is computationally hard. Approximation algorithms for Tverberg partition negatively impact the resilience guarantees of consensus algorithm. To deal with this issue, we propose to use the idea of centerpoint, which is an extension of median in higher dimensions, instead of Tverberg partition. We show that the resilience of such algorithms to adversarial nodes is improved if we use the notion of centerpoint. Furthermore, using centerpoint provides a better characterization of the necessary and sufficient conditions guaranteeing resilient vector consensus. We analyze these conditions in two, three, and higher dimensions separately. We also numerically evaluate the performance of our approach.

RSA is a popular asymmetric key algorithm with two keys scheme, a public key for encryption and private key for decryption. RSA has weaknesses in encryption and decryption of data, including slow in the process of encryption and decryption because it uses a lot of number generation. The reason is RSA algorithm can work well and is resistant to attacks such as brute force and statistical attacks. in this paper, it aims to strengthen the scheme by combining RSA with the One Time Pad algorithm so that it will bring up a new design to be used to enhance security on two-factor authentication. Contribution in this paper is to find a new scheme algorithm for an enhanced scheme of RSA. One Time Pad and RSA can combine as well.

The security of energy supply in a power grid critically depends on the ability to accurately estimate the state of the system. However, manipulated power flow measurements can potentially hide overloads and bypass the bad data detection scheme to interfere the validity of estimated states. In this paper, we use an autoencoder neural network to detect anomalous system states and investigate the impact of hyperparameters on the detection performance for false data injection attacks that target power flows. Experimental results on the IEEE 118 bus system indicate that the proposed mechanism has the ability to achieve satisfactory learning efficiency and detection accuracy.

Currently the use and production of multimedia data such as digital images have increased due to its wide use within smart devices and open networks. Although this has some advantages, it has generated several issues related to the infraction of intellectual property. Digital image watermarking is a promissory solution to solve these issues. Considering the need to develop mechanisms to improve the information security as well as protect the intellectual property of the digital images, in this paper we propose a novel visible-imperceptible watermarking based on reversible data hiding with contrast enhancement. In this way, a watermark logo is embedded in the spatial domain of the original image imperceptibly, so that the logo is revealed applying reversible data hiding increasing the contrast of the watermarked image and the same time concealing a great amount of data bits, which are extracted and the watermarked image restored to its original conditions using the reversible functionality. Experimental results show the effectiveness of the proposed algorithm. A performance comparison with the current state-of-the-art is provided.