Adversarial Model Research Articles

Toward Trustworthy Decision-Making for Autonomous Vehicles: A Robust Reinforcement Learning Approach with Safety Guarantees

While autonomous vehicles are vital components of intelligent transportation systems, ensuring the trustworthiness of decision-making remains a substantial challenge in realizing autonomous driving. Therefore, we present a novel robust reinforcement learning approach with safety guarantees to attain trustworthy decision-making for autonomous vehicles. The proposed technique ensures decision trustworthiness in terms of policy robustness and collision safety. Specifically, an adversary model is learned online to simulate the worst-case uncertainty by approximating the optimal adversarial perturbations on the observed states and environmental dynamics. In addition, an adversarial robust actor-critic algorithm is developed to enable the agent to learn robust policies against perturbations in observations and dynamics. Moreover, we devise a safety mask to guarantee the collision safety of the autonomous driving agent during both the training and testing processes using an interpretable knowledge model known as the Responsibility-Sensitive Safety Model. Finally, the proposed approach is evaluated through both simulations and experiments. These results indicate that the autonomous driving agent can make trustworthy decisions and drastically reduce the number of collisions through robust safety policies.

An efficient authentication protocol for smart grid communication based on on-chip-error-correcting physical unclonable function

Security has become a main concern for the smart grid to move from research and development to industry. Central to this security paradigm is the concept of resistance to threats, whether emanating from an active adversary seeking to disrupt operations or a passive entity covertly intercepting sensitive data. In this dynamic landscape, smart meters (SMs) stand as the sentinels at the edge of the grid, silently gathering and transmitting invaluable data. Yet, this very placement often leaves them vulnerable in unprotected areas, underscoring the paramount importance of physical security in the smart grid. It is here that Physical Unclonable Functions (PUFs) have emerged as a formidable ally. These unique constructs serve as guardians of physical security, leveraging the inherent unpredictability of manufacturing processes to create cryptographic keys. PUFs, however, are not without their own conundrums, primarily in the realm of reliability. This challenge has prevented their widespread integration into cryptographic applications. Fuzzy extractors have been considered as a solution to solve the reliability problem of PUFs, albeit at the cost of imposing significant computational burdens. To that end, we first propose an on-chip-error-correcting (OCEC) PUF that efficiently generates stable digits for the authentication process. Afterward, we introduce a lightweight authentication protocol between the SMs and neighborhood gateway (NG) based on the proposed PUF. The provable security analysis shows that not only the proposed protocol can stand secure in the Canetti–Krawczyk (CK) adversary model but also provides additional security features. Also, the performance evaluation demonstrates the significant improvement of the proposed scheme in comparison with the state-of-the-art.

A transformer-based generative adversarial learning to detect sarcasm from Bengali text with correct classification of confusing text

Sarcasm detection research in Bengali is still limited due to a lack of relevant resources. In this context, getting high-quality annotated data is costly and time-consuming. Therefore, in this paper, we present a transformer-based generative adversarial learning for sarcasm detection from Bengali text based on available limited labeled data. Here, we use the Bengali sarcasm dataset ‘Ben-Sarc’. Besides, we construct another dataset containing Bengali sarcastic and non-sarcastic comments from YouTube and newspapers to observe the model's performance on the new dataset. On top of that, we utilize another Bengali sarcasm dataset ‘BanglaSarc’ to further prove our models' robustness. Among all models, the Bangla BERT-based Generative Adversarial Model has achieved the highest accuracy with 77.1% for the ‘Ben-Sarc’ dataset. Besides, this model has achieved the highest accuracy of 68.2% for the dataset constructed from YouTube and newspaper, and 97.2% for the ‘BanglaSarc’ dataset.

MTS-DVGAN: Anomaly detection in cyber-physical systems using a dual variational generative adversarial network

Deep generative models are promising in detecting novel cyber-physical attacks, mitigating the vulnerability of Cyber-physical systems (CPSs) without relying on labeled information. Nonetheless, these generative models face challenges in identifying attack behaviors that closely resemble normal data, or deviate from the normal data distribution but are in close proximity to the manifold of the normal cluster in latent space. To tackle this problem, this article proposes a novel unsupervised dual variational generative adversarial model named MST-DVGAN, to perform anomaly detection in multivariate time series data for CPS security. The central concept is to enhance the model's discriminative capability by widening the distinction between reconstructed abnormal samples and their normal counterparts. Specifically, we propose an augmented module by imposing contrastive constraints on the reconstruction process to obtain a more compact embedding. Then, by exploiting the distribution property and modeling the normal patterns of multivariate time series, a variational autoencoder is introduced to force the generative adversarial network (GAN) to generate diverse samples. Furthermore, two augmented loss functions are designed to extract essential characteristics in a self-supervised manner through mutual guidance between the augmented samples and original samples. Finally, a specific feature center loss is introduced for the generator network to enhance its stability. Empirical experiments are conducted on three public datasets, namely SWAT, WADI and NSL_KDD. Comparing with the state-of-the-art methods, the evaluation results show that the proposed MTS-DVGAN is more stable and can achieve consistent performance improvement.

Privacy-preserving recommendation system based on user classification

Recommender systems have become ubiquitous in many application domains such as e-commerce and entertainment to recommend items that are interesting to the users. Collaborative Filtering is one of the most widely known techniques for implementing a recommender system, it models user–item interactions using data such as ratings to predict user preferences, which could potentially violate user privacy and expose sensitive data. Although there exist solutions for protecting user data in recommender systems, such as utilising cryptography, they are less practical due to computational overhead. In this paper, we propose RSUC, a privacy-preserving Recommender System based on User Classification. RSUC incorporates homomorphic encryption for better data confidentiality. To mitigate performance issues, RSUC classifies similar users in groups and computes the recommendation in a group while retaining privacy and accuracy. Furthermore, an optimised approach is applied to RSUC to further reduce communication and computational costs using data packing. Security analysis indicates that RSUC is secure under the semi-honest adversary model. Experimental results show that RSUC achieves 4× performance improvement over the standard approach and offers 54× better overall performance over the existing solution.

Enhancing DDoS Attack Detection in SDNs with GAN-Based Imbalanced Data Augmentation

Securing computer networks has become crucial due to the ongoing emergence of diverse network attacks. The popularity of Software Defined Networks (SDN) has risen because of its ability to enhance network agility, efficiency, and adaptability to recent networking challenges. However, it is essential to note that SDNs, which depend on centralized controllers, can be severely affected by Distributed Denial of Service (DDoS) attacks. The threat of DDoS attacks has grown exponentially, resulting in the evolution of robust Machine Learning-based DDoS attack detection systems within SDN. DDoS attack detection systems may deliver poor performance when trained on imbalanced datasets. Traditional techniques for handling imbalanced datasets need to be revised. Recent advances in generative adversarial networks (GANs) have revealed significant potential in generating synthetic data while preserving the probability distribution of the original data. This innovative procedure offers a promising solution to mitigate the challenges of imbalanced data in DDoS attack detection. To address challenges originating from imbalanced training datasets, we employed Generative Adversarial models to generate adversarial attacks from one viewpoint and evaluate their quality from another perspective. We chose Generative Adversarial Networks (GANs), Bidirectional GANs (Bi-GANs), and Wasserstein GANs (WGANs) based on extensive usage and reliability criteria in various domains. We conducted a comprehensive assessment to evaluate their effectiveness and resilience in generating high-quality attacks. It helps to develop, train, and fine-tune machine and deep learning models to estimate their impacts. We utilized NSL-KDD and CICIDS-2017 datasets to ensure generalization, implementing both ML and DL approaches. The outcomes demonstrate that the WGAN model outperformed GAN, Bi-GAN, and the models trained on the original imbalanced dataset and traditional sampling techniques in binary and multiclass classifications for both datasets.

Security Analysis and Enhancement of INTERBUS Protocol in ICS Based on Colored Petri Net

The integration of buses in industrial control systems, fueled by advancements such as the Internet of Things (IoT), has led to their widespread adoption, significantly enhancing operational efficiency. However, with the increasing interconnection of systems, ensuring the security of bus communications and protocols has become an urgent priority. This paper focuses on addressing the specific security concerns associated with the widely adopted INTERBUS protocol—a fieldbus protocol. Our approach leverages the theory of colored Petri nets (CPN) for modeling, enabling a comprehensive analysis of the protocol’s security. Rigorous formal verification and analysis of the security protocol are conducted by employing the Dolev–Yao adversary model. Our investigation reveals the presence of three critical vulnerabilities: replay attacks, tampering, and impersonation. To fortify the security of the protocol, we propose the introduction of a key distribution center and the utilization of hash values. Through meticulous analysis and verification, our proposed enhancements effectively reinforce the security performance of the INTERBUS protocol.

Perceptual contrast and residual self-attention generative adversarial network-based for highly under-sampled MRI reconstruction

ObjectiveThe study of deep learning-based fast magnetic resonance imaging (MRI) reconstruction methods has become popular in recent years. However, it is still challenging to obtain high-fidelity MRI images from undersampled data with large acceleration factors. The objective of this study was to improve the reconstruction quality of under-sampled MR images by exploring the correlation between pixels in the image and the perceptual similarities between the initial reconstructed images and the fully sampled ones. MethodsExisting reconstruction methods often suffer from aliasing artefacts and blurry texture information. We propose a novel perceptual residual self-attention generative adversarial model, SRSAGAN, to tackle this issue for fast and accurate MRI reconstruction. The proposed method involves two operation steps. In step 1, leveraging the average information and similarity weights of high-level features, we designed a novel feature enhancement module based on residual self-attention to restore fine texture details in the reconstructed image. In step 2, to improve the accuracy of network inference, we compare the perceptual similarities between the initial reconstructed images and the fully sampled ones. ResultThe proposed algorithm was evaluated using the fastMRI dataset. It demonstrates that SRSAGAN outperforms state-of-art methods and reconstructs complex image texture with only 10 % retained the k-space raw data (10-fold acceleration). The reconstruction time is only about 7.78 ms.

Category-level selective dual-adversarial network using significance-augmented unsupervised domain adaptation for surface defect detection

Surface defect detection is very important to ensure the quality of industrial products. Traditional machine learning cannot be well extended to a non-identically distributed dataset, making surface defect detection a data-limited task. Unsupervised domain adaptation (UDA) can solve this problem by transferring knowledge from a labeled source domain to an unlabeled target domain. Adversarial learning is one of the latest heuristic methods to deal with domain shift in UDA tasks. Although impressive results have been achieved, the adversarial model still suffers from the equilibrium challenge, which may lead to under-transfer or negative transfer. To this end, we utilize joint distribution adaptation to propose a novel UDA model, named significance-augmented-based category-level selective dual-adversarial (CDASA) network, to learn a generalized model. Specifically, to promote positive transfer, we use the selective dual-adversarial (DA) learning strategy to further minimize the feature distribution difference at equilibrium to achieve better domain confusion. Meanwhile, guided by defect recognition performance, the transferability of confusion domain is measured to enhance the distributions of potential domains. In addition, to avoid under-transfer, we consider the relationship between the target data and the decision boundary, and then the significance-augmented (SA) mechanism is proposed to encourage class-level alignment. Thus, the alignment features with domain-invariant and category discrimination can be captured simultaneously. Extensive experiments on collected real industrial datasets and publicly available steel surface defect datasets confirm the effectiveness of our approach.

Gathering Over Heterogeneous Meeting Nodes

Abstract We consider two finite and disjoint sets of homogeneous robots deployed at the nodes of an infinite grid graph. The grid graph also comprises two finite and disjoint sets of prefixed meeting nodes located over the nodes of the grid. The objective of our study is to design a distributed algorithm that gathers all the robots belonging to the first team at one of the meeting nodes belonging to the first type, and all the robots in the second team must gather at one of the meeting nodes belonging to the second type. The robots can distinguish between the two types of meeting nodes. However, a robot cannot identify its team members. This paper assumes the strongest adversarial model, namely the asynchronous scheduler. We have characterized all the initial configurations for which the gathering problem is unsolvable. For the remaining initial configurations, the paper proposes a distributed gathering algorithm. Assuming the robots are capable of global-weak multiplicity detection, the proposed algorithm solves the problem within a finite time period. The algorithm runs in $\Theta (dn)$ moves and $O(dn)$ epochs, where $d$ is the diameter of the minimum enclosing rectangle of all the robots and meeting nodes in the initial configuration, and $n$ is the total number of robots in the system.

Post‐quantum secure two‐party computing protocols against malicious adversaries

SummarySecure two‐party computation allows a pair of parties to compute a function together while keeping their inputs private. Ultimately, each party receives only its own correct output. In this paper, a post‐quantum secure two‐party computation protocol is proposed that can be used to effectively block malicious parties. The protocol solves the problems of traditional protocols based on garbled circuits, which are vulnerable to quantum attacks, high communication costs and low computational efficiency. The input garbled keys of the circuit constructor is structured as a Learning with Error (LWE) equation, enabling the circuit constructor to employ a zero‐knowledge proof that demonstrates the uniformity of inputs across all circuits.In the key transfer phase, an LWE‐based batch single‐choice cut‐and‐choose oblivious transfer is proposed to avoid selective failure attacks. In addition, the protocol employs a penalty mechanism to detect if the circuit constructor has generated an incorrect circuit. We have compared the communication overhead of this protocol with three other secure two‐party computation protocols based on Cut‐and‐Choose technology. The analytical results show that this protocol has the best error probability and is resilient to quantum attacks under the malicious adversary model. In addition, with appropriate parameters, the protocol is able to reduce its communication bandwidth by an average of 40.41%.

Proof of retrievability with flexible designated verification for cloud storage

With numerous applications in cloud storage, Proof of Retrievability (PoR) is an efficient solution for verification and retrieval of big data on a remote server. Existing PoR schemes can support two mutually exclusive verification modes, namely either private verifiability or public verifiability, depending on whether the outsourced data is verifiable by the data owner (user) only or by everyone. In order to allow the user to flexibly designate the capability of verification, we propose a novel framework for PoR schemes that enables fine-grained control of remote data verification for cloud storage. Our idea is enabling the user to dynamically release some tokens (verification keys) to one or more third parties (designated verifiers), such that only the user and the designated verifiers can verify and retrieve the outsourced data from the server. We formalize this notion as PoR with Flexible Designated Verification (PoR-FDV), and define an extended adversarial model for PoR-FDV based on Juel et al.'s PoR model. Under the new framework, we propose a generic, simple and elegant construction of PoR-FDV from public verifiable PoR and identity-based KEM/DEM, which then gives birth to a concrete and efficient PoR-FDV scheme. Compared to existing PoR schemes, PoR-FDV provides higher flexibility and scalability, as well as many useful features.

Fortified-Grid: Fortifying Smart Grids through the Integration of the Trusted Platform Module in Internet of Things Devices

This paper presents a hardware-assisted security primitive that integrates the Trusted Platform Module (TPM) into IoT devices for authentication in smart grids. Data and device security plays a pivotal role in smart grids since they are vulnerable to various attacks that could risk grid failure. The proposed Fortified-Grid security primitive provides an innovative solution, leveraging the TPM for attestation coupled with standard X.509 certificates. This methodology serves a dual purpose, ensuring the authenticity of IoT devices and upholding software integrity, an indispensable foundation for any resilient smart grid security system. TPM is a hardware security module that can generate keys and store them with encryption so they cannot be compromised. Formal security verification has been performed using the random or real Oracle (ROR) model and widely accepted AVISPA simulation tool, while informal security verification uses the DY and CK adversary model. Fortified-Grid helps to validate the attested state of IoT devices with a minimal network overhead of 1984 bits.

Adversarial Modeling Regularization for Modeling and Linearization of Power Amplifiers

Overfitting problem is very common in the coefficient identification process of digital predistortion (DPD) technique. Especially in some related DPD topics, the overfitting problem seriously limits the linearization performance. In this article, to improve the robustness of coefficient identification, we present the adversarial modeling regularization (AMR) technique. By regarding the linear-in-coefficient power amplifier (PA) behavioral models as the prior knowledge of the PA itself, the proposed regularization constrains the PA model’s coefficients by evaluating the fitting performance of other adversarial models. In addition, we present two complexity reduction strategies for the AMR. The AMR is extended and validated under four related tasks that are prone to overfitting: the few-sample learning, the piecewise DPD technique, the aliasing-sampling technique, and the band-limited technique. The experiment results show that the proposed technique can dramatically improve the forward behavioral modeling and linearization performance compared with the other state-of-the-art algorithms in all tasks.

Основные подходы к понятию и видам обвинения в российском уголовном процессе

The subject of this study is the institute of prosecution in domestic criminal proceedings. In the fi rst part of the article, the author explores the concept of accusation. Approaches are considered in which the accusation was identifi ed both with a criminal suit and with criminal prosecution. The second part of the presented work is devoted to the types of charges and the criteria for their differentiation. A summary analysis of the positions of modern, Soviet and pre-revolutionary researchers of law on this issue is given. Various types of charges are considered depending on the subject entitled to its implementation. Particular attention is paid to the so-called public prosecution, which was supported by citizens who were not endowed with this right ex offi cio. Based on the study, the author comes to the conclusion that the adversarial model of the criminal process is characterized by the presence of various types of charges, as well as various subjects entitled to its implementation. At the moment, in the Russian criminal process there is a so-called “prosecutor’s monopoly” on the prosecution, which is somewhat limited by the presence of a private prosecution. However, a small number of offenses classifi ed as private prosecution, as well as the number of criminal cases in this category considered by the courts, does not change the existing paradigm. Thus, the existing procedure does not fully correspond to the adversarial model of the criminal process.

Evaluating the impact of generative adversarial models on the performance of anomaly intrusion detection

AbstractWith the increasing rate and types of cyber attacks against information systems and communication infrastructures, many tools are needed to detect and mitigate against such attacks, for example, Intrusion Detection Systems (IDSs). Unfortunately, traditional Signature‐based IDSs (SIDSs) perform poorly against previously unseen adversarial attacks. Anomaly‐based IDSs (AIDSs) use Machine Learning (ML) and Deep Learning (DL) approaches to overcome these limitations. However, AIDS performance can be poor when trained on imbalanced datasets. To address the challenge of AIDS performance caused by these unbalanced training datasets, generative adversarial models are proposed to obtain adversarial attacks from one side and analyse their quality from another. According to extensive usage and reliability criteria for generative adversarial models in different disciplines, Generative Adversarial Networks (GANs), Bidirectional GAN (BiGAN), and Wasserstein GAN (WGAN) are employed to serve AIDS. The authors have extensively assessed their abilities and robustness to deliver high‐quality attacks for AIDS. AIDSs are constructed, trained, and tuned based on these models to measure their impacts. The authors have employed two datasets: NSL‐KDD and CICIDS‐2017 for generalisation purposes, where ML and DL approaches are utilised to implement AIDSs. Their results show that the WGAN model outperformed GANs and BiGAN models in binary and multiclass classifications for both datasets.

Post-Quantum Adversarial Modeling: A User’s Perspective

Post-Quantum Adversarial Modeling: A User’s Perspective

Universal Adversarial Training Using Auxiliary Conditional Generative Model-Based Adversarial Attack Generation

While Machine Learning has become the holy grail of modern-day computing, it has many security flaws that have yet to be addressed and resolved. Adversarial attacks are one of these security flaws, in which an attacker appends noise to data samples that machine learning models take as input with the aim of fooling the model. Various adversarial training methods have been proposed that augment adversarial examples in the training dataset for defense against such attacks. However, a general limitation exists where a robust model can only protect itself against adversarial attacks that are known or similar to those it was trained on. To address this limitation, this paper proposes a Universal Adversarial Training algorithm using adversarial examples generated by an Auxiliary Classifier Generative Adversarial Network (AC-GAN) in parallel with other data augmentation techniques, such as the mixup method. This method builds on a previously proposed technique, Adversarial Training, in which adversarial examples produced by gradient-based methods are augmented and added to the training data. Our method improves the AC-GAN architecture for adversarial example generation to make it more suitable for adversarial training by updating different loss terms and testing its performance against various attacks compared to other robust adversarial models. In this way, it becomes apparent that generative models are better suited for boosting adversarial robustness through adversarial training. When tested using various attack types, our proposed model had an average accuracy of 97.48% on the MNIST dataset and 94.02% on the CelebA dataset, proving that generative models have a higher chance of boosting adversarial security through adversarial training.

A histogram-driven generative adversarial network for brain MRI to CT synthesis

Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) are commonly used tools for medical diagnostic assessment. Considering the ionizing radiation of CT imaging, estimating CT images from radiation-free MRI would be beneficial for medical diagnosis. Although state-of-art generative adversarial networks or end-to-end architecture models can generate realistic natural images, it is challenging to generate medical CT images with high signal-to-noise ratios and are paired with MRI. We propose a histogram-driven generative adversarial network (HisGAN) to address this issue, estimate CT images paired with MR, and develop a histogram-based dynamic scaling factor to facilitate learning different image styles. By employing an adversarial learning strategy to train the end-to-end generator, the generator better simulate the nonlinear mapping from source to target. For the generator, the proposed method applies multiple learnable parameters to adjust the overall weights of the dilated convolution layers to ensure sufficient expansive receptive fields for improved performance. Additionally, the method utilizes deep residual networks to train randomly smoothed generated images and employs adversarial loss to enhance the generation of the discriminator, achieving a balance between the generator and the discriminator. Our approach can synthesize image details at the pixel level in the target domain and has been evaluated using two datasets for MR to CT, T1 to T2, and Flair to T1ce modality synthesis tasks. The proposed method outperforms existing generative adversarial models applied to medical image synthesis.

Federated Learning of Generative Image Priors for MRI Reconstruction.

Multi-institutional efforts can facilitate training of deep MRI reconstruction models, albeit privacy risks arise during cross-site sharing of imaging data. Federated learning (FL) has recently been introduced to address privacy concerns by enabling distributed training without transfer of imaging data. Existing FL methods employ conditional reconstruction models to map from undersampled to fully-sampled acquisitions via explicit knowledge of the accelerated imaging operator. Since conditional models generalize poorly across different acceleration rates or sampling densities, imaging operators must be fixed between training and testing, and they are typically matched across sites. To improve patient privacy, performance and flexibility in multi-site collaborations, here we introduce Federated learning of Generative IMage Priors (FedGIMP) for MRI reconstruction. FedGIMP leverages a two-stage approach: cross-site learning of a generative MRI prior, and prior adaptation following injection of the imaging operator. The global MRI prior is learned via an unconditional adversarial model that synthesizes high-quality MR images based on latent variables. A novel mapper subnetwork produces site-specific latents to maintain specificity in the prior. During inference, the prior is first combined with subject-specific imaging operators to enable reconstruction, and it is then adapted to individual cross-sections by minimizing a data-consistency loss. Comprehensive experiments on multi-institutional datasets clearly demonstrate enhanced performance of FedGIMP against both centralized and FL methods based on conditional models.