Adversarial Effects Research Articles

Multiproduct firm’s reputation and leniency program in multimarket collusion

The corporate leniency program is believed to be useful for antitrust authorities. Our study challenges this belief in a setting where a large multiproduct firm meets single-product firms in independent and isolated markets. We explore an adversary effect of the corporate leniency program in multimarket collusion in a reputation model. A multiproduct firm forming cartels in multiple markets can build its reputation as a tough firm that punishes any deviation by applying for leniency. We show that the multiproduct firm can manipulate the leniency program to stabilize cartels in markets without material linkages (such as demand linkages). This effect does not exist if only one market exists or the leniency-application outcome is not publicly observable. Our findings theoretically explain why the numerous leniency applications by multiproduct firms should concern antitrust agencies and imply several directions for revising the leniency policy.

Defed: An Edge-Feature-Enhanced Image Denoised Network Against Adversarial Attacks for Secure Internet of Things

With the prosperous development of Internet of Things (IoT), IoT devices have been deployed in various applications, which generates large volume of image data to trace and record the users’ behaviors, resulting in better IoT services. To accurately analyze these huge data to further improve users’ experience on IoT services, deep neural networks (DNNs) are gaining more attention and have become increasingly popular. However, recent studies have shown that DNN models are vulnerable to adversarial attacks, which leads to the risk of applications in practice. Previous works are devoted to extract invariant features from the content circled by edges in images, while such features can not efficiently deal with the adversarial effect. In this work, we first study this problem from a new angle by exploring the edge feature information, which is intractable to be influenced by adversarial attacks demonstrated by our empirical analysis. Based on this, we propose a novel edge feature enhanced defense approach called Defed which incorporates edge feature information into denoised network to defend against various adversarial attacks in image area. For training phase, we only add benign images as the input and exert Gaussian noise to substitute the adversarial attacks to mitigate the dependency of models on specific adversarial attacks. For inference, we design a combination of multiple Defeds trained by different Gaussian noise levels and deploy confidence intervals to judge whether an image is adversarial or not. Experiments over real-world datasets on image classification demonstrate the efficacy and superiority compare to the state-of-the-art defense approaches.

Exogenous application of selenium and nano-selenium alleviates salt stress and improves secondary metabolites in lemon verbena under salinity stress

Salinity stress is one of the most serious abiotic factors that affects the growth, performance, and secondary compounds of medicinal plants. The aim of this study was to determine the effect of foliar application of selenium and nano-selenium separately on growth, essential oils, physiological parameters, and some secondary metabolites of Lemon verbena under salinity stress. The results showed that selenium and nano-selenium significantly increased growth parameters, photosynthetic pigments, and relative water content. Compared to the control, a higher accumulation of osmolytes (i.e., proline, soluble sugars, and total protein) and higher antioxidant activity were observed in the selenium-treated plants. In addition, selenium alleviated the adversary effect of oxidative stress, resulting from salinity, by reducing leaf electrolyte leakage, malondialdehyde, and H2O2 accumulation. Furthermore, selenium and nano-selenium enhanced the biosynthesis of secondary metabolites such as essential oils, total phenolic content, and flavonoid compounds under non-stress and salinity conditions. They also reduced Na+ accumulation in the root and shoot of the salinity-treated plants. Hence it can be concluded that exogenous application of selenium and nano-selenium separately can mitigate the negative effects of salinity by improving the quantitative and qualitative performance of lemon verbena plants under salinity stress.

A Blockchain-Based Data-Driven Fault-Tolerant Control System for Smart Factories in Industry 4.0

Modern technologies and data-driven approaches have enabled fault-tolerant controllers in Industry 4.0 smart factories to detect, identify, and mitigate anomalies in real-time with a high level of accuracy. However, this has also presented new challenges and requirements for cybersecurity, data analytics, and computational complexity for Data-Driven Fault-Tolerant Controllers (DD-FTC) in smart factories. To address these issues, a Blockchain-Based Data-Driven Fault-Tolerant Control (BB-DD-FTC) framework for smart factories is proposed in this paper. Blockchain ensures the integrity of data logs via its immutable ledger and decentralized architecture. Moreover, the blockchain smart contract functionality, embedded with a Data-Driven Intrusion Detection System (DD-IDS), and reconfiguration conditions, realizes DD-FTC and undertakes the mitigation response in case of cyber-attacks. The DD-IDS mechanism utilizes the principal component analysis technique and observer models, trained via neural networks, to detect an attack and identify the compromised component. The Tennessee Eastman (TE) industrial benchmark process is considered a case study to investigate the performance of the proposed framework. Two kinds of integrity attacks are applied to the sensors of the TE process with simulation results demonstrating the effectiveness of the method in mitigating the adversarial effect of the applied attacks on the overall system performance. As feedback delays can negatively impact performance, a detailed delay analysis is performed using network calculus. The security advantages and limitations of the proposed method are finally highlighted in the performed security analysis. The results are encouraging for the wider adoption of the control-over-the-blockchain concept.

3D adversarial attacks beyond point cloud

Recently, 3D deep learning models have been shown to be susceptible to adversarial attacks like their 2D counterparts. Most of the state-of-the-art (SOTA) 3D adversarial attacks perform perturbation to 3D point clouds. To reproduce these attacks in the physical scenario, a generated adversarial 3D point cloud needs to be reconstructed to mesh, which leads to a significant drop in its adversarial effect. In this paper, we propose a strong 3D adversarial attack named Mesh Attack to address this problem by directly performing perturbation on mesh of a 3D object. In order to take advantage of the most effective gradient-based attack, a differentiable sample module that back-propagate the gradient of point cloud to mesh is introduced. To further ensure the adversarial mesh examples without outlier and 3D printable, three mesh losses are adopted. Extensive experiments demonstrate that the proposed scheme outperforms SOTA 3D attacks by a significant margin. We also achieved SOTA performance under various defenses. Our code will be available at: https://github.com/cuge1995/Mesh-Attack.

BESRGAN: Boundary equilibrium face super‐resolution generative adversarial networks

AbstractExisting Generative Adversarial Networks (GAN)‐based face hallucination algorithms are hard to control the face fidelity of the generated samples, and easily generate flawed faces with unfavourable artefacts and distortions. To address this problem, the authors propose a fidelity‐controllable face super‐resolution (FSR) network boundary equilibrium face super‐resolution generative adversarial networks (BESRGAN), a fidelity ratio is introduced in their network to control how much the adversarial effect the discriminator is put on the generator; therefore, the authors’ network better trades off the objective and perceptual quality. Additionally, the authors design an equilibrium perceptual discriminator to match the perception loss distributions. Under the equilibrium constraint, the discriminator pays more attention to learning fine‐grained feature statistics of ground truths, and further guides the generator to produce photo‐realistic faces, especially in terms of facial textures. Moreover, the authors propose a novel channel‐spatial attention module (CSAM) to eliminate local distortions, by further fusing richer information from the facial prior knowledge and global high‐level facial descriptions. Extensive experiments illustrate that the authors’ approach preserves high pixel‐wise accuracy while achieving superior visual performance against state‐of‐the‐art methods. Specifically, the peak signal to noise ratio (PSNR) and structural similarity index (SSIM) of the authors’ proposed BESRGAN rise 0.64 dB and 0.02 for CelebA compared with one of the state‐of‐the‐art face super‐resolution (FSR) methods.

EITGAN: A Transformation-based Network for recovering adversarial examples

<abstract><p>Adversarial examples have been shown to easily mislead neural networks, and many strategies have been proposed to defend them. To address the problem that most transformation-based defense strategies will degrade the accuracy of clean images, we proposed an Enhanced Image Transformation Generative Adversarial Network (EITGAN). Positive perturbations were employed in the EITGAN to counteract adversarial effects while enhancing the classified performance of the samples. We also used the image super-resolution method to mitigate the effect of adversarial perturbations. The proposed method does not require modification or retraining of the classifier. Extensive experiments demonstrated that the enhanced samples generated by the EITGAN effectively defended against adversarial attacks without compromising human visual recognition, and their classification performance was superior to that of clean images.</p></abstract>

A Novel MegaBAT Optimized Intelligent Intrusion Detection System in Wireless Sensor Networks

Wireless Sensor Network (WSN), which finds as one of the major components of modern electronic and wireless systems. A WSN consists of numerous sensor nodes for the discovery of sensor networks to leverage features like data sensing, data processing, and communication. In the field of medical health care, these network plays a very vital role in transmitting highly sensitive data from different geographic regions and collecting this information by the respective network. But the fear of different attacks on health care data typically increases day by day. In a very short period, these attacks may cause adversarial effects to the WSN nodes. Furthermore, the existing Intrusion Detection System (IDS) suffers from the drawbacks of limited resources, low detection rate, and high computational overhead and also increases the false alarm rates in detecting the different attacks. Given the above-mentioned problems, this paper proposes the novel MegaBAT optimized Long Short Term Memory (MBOLT)-IDS for WSNs for the effective detection of different attacks. In the proposed framework, hyperparameters of deep Long Short-Term Memory (LSTM) were optimized by the meta-heuristic megabat algorithm to obtain a low computational overhead and high performance. The experimentations have been carried out using (Wireless Sensor NetworkDetection System)WSN-DS datasets and performance metrics such as accuracy, recall, precision, specificity, and F1-score are calculated and compared with the other existing intelligent IDS. The proposed framework provides outstanding results in detecting the black hole, gray hole, scheduling, flooding attacks and significantly reduces the time complexity, which makes this system suitable for resource-constraint WSNs.

Hamka’s Daʿwah Reform: Islamisation of Self, Education and Institution

Abdul Malik Karim Amrullah or HAMKA (1908-1981) was a prominent Malay scholar and also an icon of daʿwah Islamiyyah. Although his name and works are common to the Malays, his daʿwah orientation needs to be explored further in view of the lethargic conditions of present daʿwah movements in the Malay lands. Granted the title Pahlawan Nasional Indonesia, Hamka’s thought epitomised the idealism of daʿwah reform, and his activism symbolises the institutional configuration of daʿwah. Hamka went beyond the rhetoric slogan of Islam as a way of life when he brought tangible solutions to the problems that happen in society. The core inventory to that tangible solution, in his opinion, should start with oneself i.e. Islamisation of self, before any meaningful daʿwah can be administered to the community and institutions. This article attempts to promote Hamka’s daʿwah reform; a synergised intellectualism and activism that are missing from the contemporary da’wah orientations and led to its adversary effect on public understanding and practice of Islam.

Master equation emulation and coherence preservation with classical control of a superconducting qubit

Open quantum systems are a topic of intense theoretical research. The use of master equations to model a system's evolution subject to an interaction with an external environment is one of the most successful theoretical paradigms. General experimental tools to study different open system realizations have been limited, and so it is highly desirable to develop experimental tools which emulate diverse master equation dynamics and give a way to test open systems theories. In this paper we demonstrate a systematic method for engineering specific system-environment interactions and emulating master equations of a particular form using classical stochastic noise in a superconducting transmon qubit. We also demonstrate that non-Markovian noise can be used as a resource to extend the coherence of a quantum system and counteract the adversarial effects of Markovian environments.

Physical layer security using boundary technique for emerging wireless communication systems

AbstractThe broadcast nature of radio propagation in wireless communication has been suspected as the loopholes of passive or active attacks by unauthorized users (eavesdroppers). The physical layer security techniques operate at the lowest stack of OSI layer against conventional cryptographic approaches, operating at the upper layer. However, techniques such as channel coding, power (directional antenna and artificial noise), and spread spectrum have been (and continuously) deployed to safeguard against sophisticated attacks. Most of these deployments are theoretical, and a few are enhanced for efficient security against an intruder. In this article, a boundary technique approach is proposed and applied to the physical layer to improve its secrecy‐capacity and subdue adversary effects at the legitimate receiver. Hybrid performance metrics were adopted, and a Monte Carlo simulation was performed. The result obtained using secrecy outage probability, secrecy‐capacity, and intercept‐probability show that our proposed techniques enhance the secret transmission between the main transmitter and legitimate receiver. The simulation results were compared with the analytical methods. It was found that the channel between the transmitter and the main receiver has a better signal‐to‐noise ratio than the corresponding eavesdropper's channel. Conclusively, performance of the proposed technique is validated for applications in emerging wireless communication systems.

Witness or Interpreter?

This study demonstrates that self-interpretation does indeed occur in Kenyan courtroom proceedings, a situation that necessitates the use of a language other than the regular and official languages of Kenyan courts. Such language use rendered mostly in terms of self-interpretation has far-reaching ramifications on the content, facts, style, and meaning predispositions of a witness's testimony. Most studies in courtroom communicative interactions, language, and speech manifestations, have been largely and dominantly sociolinguistic in approach and there is so much that has been asserted in that dimension. However, this study's point of departure is that it seeks to adopt a Translation Studies approach to analyze self-interpreted presentations made by four witnesses in selected criminal cases at Kisii Law Courts in Kenya. Their self-interpreted testimonies which constitute the data used in this discussion were collected between October 2020 and June 2021. The testimonies rendered constituted: one murder case in the high court, one rape case, and two assault cases in the magistrate court. The overarching aim of this study is an attempt to show that, bilingual litigants, who have habitually and for a long time been regarded as persons of limited language competence, can in practical renditions be astute self-interpreting persons in testimony presentation. However, the confrontational experience they undergo throughout has adversarial effects on the facts of the case and the eventual outcomes of such cases, the disadvantages of their competence in L2 (the official language of the court) notwithstanding. Consequently, a translation studies approach, as applied in this paper, offers a framework of reference through which it is possible to analyze the encumbrances of comprehending legal procedures, terminology, and propriety which litigants undergo to accentuate meaning shifts, stem contextual meaning deviations besides the overall factual misrepresentations which emerge during and as a result of self-interpreted renditions as constrained by the contextual imperatives of traditional courtroom language.

On Supervised Class-Imbalanced Learning: An Updated Perspective and Some Key Challenges

The problem of class imbalance has always been considered as a significant challenge to traditional machine learning and the emerging deep learning research communities. A classification problem can be considered as class imbalanced if the training set does not contain an equal number of labeled examples from all the classes. A classifier trained on such an imbalanced training set is likely to favor those classes containing a larger number of training examples than the others. Unfortunately, the classes that contain a small number of labelled instances usually correspond to rare and significant events. Thus, poor classification accuracy on these classes may lead to severe consequences. In this article, we aim to provide a comprehensive summary of the rich pool of research works attempting to combat the adversarial effects of class imbalance efficiently. Specifically, following a formal definition of the problem of class imbalance, we explore the plethora of traditional machine learning approaches aiming to mitigate its adversarial effects. We further discuss the state-of-the-art deep-learning-based approaches for improving a classifier’s resilience against class imbalance and highlight the need for techniques tailored for such a paradigm. Moreover, we look at the emerging applications where class imbalance can be a major concern. Finally, we outline a few open problems along with the various challenges emerging with the advent of modern applications, deep learning paradigm, and new sources of data.

Video Performances and Interactive Features as Contributors to Tik Tok’s Online Environment

With the rising concern of Tik Tok videos propagating misinformation to younger viewers, research on the adversary effects of the platform of Tik Tok has been few and far between. This research applies a simplistic approach of gathering and observing numbers related to 60 content creators on YouTube and Tik Tok (30 on each platform) for the comparative analysis. The analysis is largely built upon the average video length of each creator’s most recent 50 videos and the frequency of numbers of lines of each top comment. The correlation between the two is found relatively weak within one platform. However, videos on YouTube have significant lengthier top comments than those on Tik Tok most of the time, which suggests a higher level of thinking and commitment on the former in general. The two factors along with different features on the two platforms that are tailored towards different types of audiences contribute to the levels of discussion and the ways users treat information and misinformation. Implications and recommendations in regard to Tik Tok’s environment are further discussed.

Adversarial Patch Attack on Multi-Scale Object Detection for UAV Remote Sensing Images

Although deep learning has received extensive attention and achieved excellent performance in various scenarios, it suffers from adversarial examples to some extent. In particular, physical attack poses a greater threat than digital attack. However, existing research has paid less attention to the physical attack of object detection in UAV remote sensing images (RSIs). In this work, we carefully analyze the universal adversarial patch attack for multi-scale objects in the field of remote sensing. There are two challenges faced by an adversarial attack in RSIs. On one hand, the number of objects in remote sensing images is more than that of natural images. Therefore, it is difficult for an adversarial patch to show an adversarial effect on all objects when attacking a detector of RSIs. On the other hand, the wide height range of the photography platform causes the size of objects to vary a great deal, which presents challenges for the generation of universal adversarial perturbation for multi-scale objects. To this end, we propose an adversarial attack method of object detection for remote sensing data. One of the key ideas of the proposed method is the novel optimization of the adversarial patch. We aim to attack as many objects as possible by formulating a joint optimization problem. Furthermore, we raise the scale factor to generate a universal adversarial patch that adapts to multi-scale objects, which ensures that the adversarial patch is valid for multi-scale objects in the real world. Extensive experiments demonstrate the superiority of our method against state-of-the-art methods on YOLO-v3 and YOLO-v5. In addition, we also validate the effectiveness of our method in real-world applications.

In vitro and in silico characteristics of doxorubicin-loaded four polymeric-based polysaccharides-modified super paramagnetic iron oxide nanoparticles for cancer chemotherapy and magnetic resonance imaging

Super Paramagnetic Iron Oxide Nanoparticles have attracted much attention in drug delivery systems because of their considerable surface-to-volume ratio and content as well. In this regard, in this study, four different formulations of magnetic nanoparticles were synthesized via various polymeric coating layer (beta-cyclodextrin, Chitosan, Gum Arabic, and Alginate) in order to find an appropriate polymer in terms of higher drug loading efficiency and capacity. Molecular dynamics simulation and Quantum mechanics were manipulated to examine the interaction mechanism between the polysaccharide polymers and Doxorubicin. Based on the calculated results, As the best candidate, βCD was chosen for DOX loading. DOX, a hydrophilic drug, was loaded into each of the four formulations. The optimum formulation, Fe3O4-ßCD, possessed loading efficiency and capacity, with 97% and 13%, respectively. The selected nanocarrier was fully characterized by Fourier Transform Infrared, VSM, TGA, XRD, SEM, UV-Vis, DLS, and TEM. TEM result exhibited the prepared Fe3O4-ßCD-DOX nanoparticles had a spherical shape (core and shell) and the size of around 15 nm. In vitro release investigation was conducted using HPLC analysis in the phosphate-buffered saline in acidic (pH = 5.4) and neutral (pH = 7.4) medium. The obtained results demonstrated that a sustain release performance for more than 72 h with the amount of 50% DOX release in pH = 5.4. Cytotoxicity assay showed the toxicity of bare Fe3O4 and Fe3O4-ßCDNPsagainst MDA-MB468 cancerous cell and MCF10A normal cell was negligible and DOX-loaded NPs (Fe3O4-ßCD-DOX) illustrated a higher toxicity in MDA-MB468 cancerous cells than MCF 10 A normal cells. The results showed that Fe3O4-ßCD-DOXNPs could improve the therapeutic efficacy and also reduce the adversarial effects; accordingly, the prepared system could be served as a promising drug delivery system in cancer treatment.

Multiple Adversarial Domains Adaptation Approach for Mitigating Adversarial Attacks Effects

Although neural networks are near achieving performance similar to humans in many tasks, they are susceptible to adversarial attacks in the form of a small, intentionally designed perturbation, which could lead to misclassifications. The best defense against these attacks, so far, is adversarial training (AT), which improves a model’s robustness by augmenting the training data with adversarial examples. However, AT usually decreases the model’s accuracy on clean samples and could overfit to a specific attack, inhibiting its ability to generalize to new attacks. In this paper, we investigate the usage of domain adaptation to enhance AT’s performance. We propose a novel multiple adversarial domain adaptation (MADA) method, which looks at this problem as a domain adaptation task to discover robust features. Specifically, we use adversarial learning to learn features that are domain-invariant between multiple adversarial domains and the clean domain. We evaluated MADA on MNIST and CIFAR-10 datasets with multiple adversarial attacks during training and testing. The results of our experiments show that MADA is superior to AT on adversarial samples by about 4% on average and on clean samples by about 1% on average.

Tekrarlayan Sinir Ağları Temelli Rüzgâr Hızı Tahmin Modelleri: Yalova Bölgesinde Bir Uygulama

Global warming and other adversarial effects caused by fossil fuel sources, renewable energy sources have been attracted more than ever. Especially, parties of Paris Climate Agreement countries pledge to reduce greenhouse gas emissions. Among renewable energy sources, wind energy is one of the significant and eligible source to produce energy sustainably. Wind energy is also one of the most important renewable energy source due to Turkey’s notable wind energy potential. Although wind energy is one of the most important clean energy sources, there are several challenges, such as intermittent and uncertain nature of wind places. Therefore, efficient and reliable energy planning and distribution mostly rely on prediction of wind energy with high accuracy. In this study, we propose four Reccurent Neural Network (RNN) methods to predict short-term wind energy production. We utilize data obtained from a station located in Yalova, Turkey to assess the performance of proposed algorithms. In our analysis, we plan to improve maintenance planning and intervene the sudden breakdowns by predicting 1 hour ahead energy production. First, we analyze the data received from the station, and the data sets were made suitable for the models. The performance results obtained from the models are plausible. Our results indicate that RNN methods can be successfully used to predict wind speed.

DEFEAT: Decoupled feature attack across deep neural networks

Adversarial attacks pose a security challenge for deep neural networks, motivating researchers to build various defense methods. Consequently, the performance of black-box attacks turns down under defense scenarios. A significant observation is that some feature-level attacks achieve an excellent success rate to fool undefended models, while their transferability is severely degraded when encountering defenses, which give a false sense of security. In this paper, we explain one possible reason caused this phenomenon is the domain-overfitting effect, which degrades the capabilities of feature perturbed images and makes them hardly fool adversarially trained defenses. To this end, we study a novel feature-level method, referred to as Decoupled Feature Attack (DEFEAT). Unlike the current attacks that use a round-robin procedure to estimate gradient estimation and update perturbation, DEFEAT decouples adversarial example generation from the optimization process. In the first stage, DEFEAT learns an distribution full of perturbations with high adversarial effects. And it then iteratively samples the noises from learned distribution to assemble adversarial examples. On top of that, we can apply transformations of existing methods into the DEFEAT framework to produce more robust perturbations. We also provide insights into the relationship between transferability and latent features that helps the community to understand the intrinsic mechanism of adversarial attacks. Extensive experiments evaluated on a variety of black-box models suggest the superiority of DEFEAT, i.e., our method fools defenses at an average success rate of 88.4%, remarkably outperforming state-of-the-art transferable attacks by a large margin of 11.5%. The code is publicly available at https://github.com/mesunhlf/DEFEAT.

Artificial Ecosystem-Based Optimization with an Improved Deep Learning Model for IoT-Assisted Sustainable Waste Management

Increasing waste generation has become a key challenge around the world due to the dramatic expansion in industrialization and urbanization. This study focuses on providing effective solutions for real-time monitoring garbage collection systems via the Internet of things (IoT). It is limited to controlling the bad odor of blowout gases and the spreading of overspills by using an IoT-based solution. The inadequate and poor dumping of waste produces radiation and toxic gases in the environment, creating an adversarial effect on global warming, human health, and the greenhouse system. The IoT and deep learning (DL) confer active solutions for real-time data monitoring and classification, correspondingly. Therefore, this paper presents an artificial ecosystem-based optimization with an improved deep learning model for IoT-assisted sustainable waste management, called the AEOIDL-SWM technique. The presented AEOIDL-SWM technique exploits IoT-based camera sensors for collecting information and a microcontroller for processing the data. For waste classification, the presented AEOIDL-SWM technique applies an improved residual network (ResNet) model-based feature extractor with an AEO-based hyperparameter optimizer. Finally, the sparse autoencoder (SAE) algorithm is exploited for waste classification. To depict the enhancements of the AEOIDL-SWM system, a widespread simulation investigation is performed. The comparative analysis shows the enhanced outcomes of the AEOIDL-SWM technique over other DL models.