Bidirectional Generative Adversarial Network Research Articles

Maize Leaf Disease Image Enhancement Algorithm Using TFEGAN

Maize leaf diseases have a profound impact on maize crop yield. However, due to practical constraints, gathering a substantial number of maize disease images is time-consuming and labor-intensive. Furthermore, low-light conditions and noise interference in maize leaf disease images present significant challenges for image generation. Therefore, this paper proposes TFEGAN, a novel algorithm designed to generate maize leaf disease images. TFEGAN leverages a Two-Way Extraction Module (TWEM) to enhance the perceptual domain of the model and effectively reuse maize leaf disease features. The algorithm employs two well-trained Generative Adversarial Networks (GANs) to enhance low-light and noisy maize leaf disease images respectively. The low-light enhancement component utilizes a double discriminator and self-regularization to balance global and local enhancement while constraining the distance between the input image and the denoising component. Moreover, the denoising part employs a strong coordinated attention mechanism and an environmental encoder to learn noise regions and their surrounding structures, thus generating better local images and enabling more accurate evaluation by the discriminant network. The generators of both networks are encapsulated by two discriminators and combined with a bi-directional feature extraction structure to generate normal maize leaf disease images. Experimental evaluations demonstrate the effectiveness of TFEGAN in improving the quality of maize leaf disease images under low-light and noisy conditions, showcasing its potential for practical applications. Overall, this paper presents a promising solution to the challenges of low-light and noise interference in maize leaf disease images, utilizing bidirectional feature extraction and GANs.

Bayesian imaging inverse problem with SA-Roundtrip prior via HMC-pCN sampler

Bayesian inference with deep generative prior has received considerable interest for solving imaging inverse problems in many scientific and engineering fields. The selection of the prior distribution is learned from, and therefore an important representation learning of, available prior measurements. The SA-Roundtrip, a novel deep generative prior, is introduced to enable controlled sampling generation and identify the data's intrinsic dimension. This prior incorporates a self-attention structure within a bidirectional generative adversarial network. Subsequently, Bayesian inference is applied to the posterior distribution in the low-dimensional latent space using the Hamiltonian Monte Carlo with preconditioned Crank-Nicolson (HMC-pCN) algorithm, which is proven to be ergodic under specific conditions. Experiments conducted on computed tomography (CT) reconstruction with the MNIST and TomoPhantom datasets reveal that the proposed method outperforms state-of-the-art comparisons, consistently yielding a robust and superior point estimator along with precise uncertainty quantification.

Learning with limited target data to detect cells in cross-modality images.

Deep neural networks have achieved excellent cell or nucleus quantification performance in microscopy images, but they often suffer from performance degradation when applied to cross-modality imaging data. Unsupervised domain adaptation (UDA) based on generative adversarial networks (GANs) has recently improved the performance of cross-modality medical image quantification. However, current GAN-based UDA methods typically require abundant target data for model training, which is often very expensive or even impossible to obtain for real applications. In this paper, we study a more realistic yet challenging UDA situation, where (unlabeled) target training data is limited and previous work seldom delves into cell identification. We first enhance a dual GAN with task-specific modeling, which provides additional supervision signals to assist with generator learning. We explore both single-directional and bidirectional task-augmented GANs for domain adaptation. Then, we further improve the GAN by introducing a differentiable, stochastic data augmentation module to explicitly reduce discriminator overfitting. We examine source-, target-, and dual-domain data augmentation for GAN enhancement, as well as joint task and data augmentation in a unified GAN-based UDA framework. We evaluate the framework for cell detection on multiple public and in-house microscopy image datasets, which are acquired with different imaging modalities, staining protocols and/or tissue preparations. The experiments demonstrate that our method significantly boosts performance when compared with the reference baseline, and it is superior to or on par with fully supervised models that are trained with real target annotations. In addition, our method outperforms recent state-of-the-art UDA approaches by a large margin on different datasets.

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.

Detection of obfuscated Tor traffic based on bidirectional generative adversarial networks and vision transform

The Onion Router (TOR) network is a decentralized system of volunteer-run servers that aims to protect the anonymity and privacy of users by routing their internet traffic through a series of nodes. Individuals who use the TOR network may employ obfuscated traffic to conceal their internet activity from network administrators or security systems attempting to block or monitor them. Furthermore, some may use obfuscated Tor traffic to hide illegal activities, such as buying and selling illegal goods or accessing illegal services on the dark web. Despite efforts to identify and block Tor traffic, challenges remain, such as a limited set of features for identification, leading to false positives and negatives. To address these challenges, this paper proposes a novel approach using Visual Transformation (ViT), and augmentation by Bidirectional Generative Adversarial Networks (BiGAN). The proposed approach demonstrates superior performance on the ISCX-Tor2016 dataset, achieving 99.59% accuracy, 99.83% recall, 99.72% precision, and 99.78% F-score, thereby outperforming current state-of-the-art techniques.

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.

An improved bidirectional generative adversarial network model for multivariate estimation of correlated and imbalanced tunnel construction parameters

An improved bidirectional generative adversarial network model for multivariate estimation of correlated and imbalanced tunnel construction parameters

Accelerated Entropic Path Sampling with a Bidirectional Generative Adversarial Network.

The role of entropy in mediating the dynamic outcomes of chemical reactions remains largely unknown. To evaluate the change of entropy along post-transition state paths, we have previously developed entropic path sampling that computes configurational entropy from an ensemble of reaction trajectories. However, one major caveat of this approach lies in its high computational demand: about 2000 trajectories are needed to converge the computation of an entropic profile. Here, by leveraging a deep generative model, we developed an accelerated entropic path sampling approach that evaluates entropic profiles using merely a few hundred reaction dynamic trajectories. The new method, called bidirectional generative adversarial network-entropic path sampling, can enhance the estimation of probability density functions of molecular configurations by generating pseudo-molecular configurations that are statistically indistinguishable from the true data. The method was established using cyclopentadiene dimerization, in which we reproduced the reference entropic profiles (derived from 2480 trajectories) using merely 124 trajectories. The method was further benchmarked using three reactions with symmetric post-transition-state bifurcation, including endo-butadiene dimerization, 5-fluoro-1,3-cyclopentadiene dimerization, and 5-methyl-1,3-cyclopentadiene dimerization. The results indicate the existence of a "hidden entropic intermediate", which is a dynamic species that binds to a local entropic maximum where no free energy minimum is formed.

Enhancing diversity in GANs via non-uniform sampling

Recent advances in Generative Adversarial Networks (GANs) have led to impressive results in generating realistic data. However, GANs training is still challenging, often leading to mode-collapse, where a certain type of samples dominates the generated output. To address this issue, we propose a novel training algorithm based on bidirectional GANs (BiGANs) that can be generalized to any implicit generative model. Our algorithm relies on a non-uniform sampling scheme, where data points in a minibatch are sampled with probability inversely proportional to their log-evidence. However, estimating log-evidence is computationally expensive. Instead, we propose to use the reconstruction error, which directly correlates with the log-evidence and only requires a BiGAN network evaluation. Additionally, we combine the aforementioned method with a regularization in the empirical distribution of the encoder that further boosts the performance. Our empirical results show that the proposed methods improve both the quality and diversity of the generated samples.

Scalable anomaly-based intrusion detection for secure Internet of Things using generative adversarial networks in fog environment

The data generated exponentially by a massive number of devices in the Internet of Things (IoT) are extremely high-dimensional, large-scale, non-labeled, which poses great challenges to timely analysis and effective decision making for anomaly detection in IoT. In this paper, we propose a novel unsupervised deep learning method to identify anomalies against IoT networks, which exploits Bidirectional Generative Adversarial Networks (BiGAN) to build model on normal IoT data. The model introduces Wasserstein distance to capture and learn the distribution of high-dimensional raw data and focuses on latent representations using an auxiliary classifier. A cycle consistency connection between data is designed to prevent information loss that helps to reduce false positive rate. The model detects outliers by utilizing reconstruction error in feature space. Another challenge facing the current anomaly detection solutions is their limited scalability, which restricts capability in handling big IoT data. This issue is resolved by deploying and jointly training the proposed method in a fog computing environment. The anomaly-based intrusion detection can be scalable by leveraging the flexibility of fog computing, which contributes to supporting efficient detection. Experimental results on two recent datasets (i.e., UNSW-NB15 and CIC-IDS2017) validate that the proposed method achieves 4% increase in accuracy and 4% reduction in false alarm rate than the state-of-the-art methods while keeping computational efficiency.

Dual attention bidirectional generative adversarial network for dynamic uncertainty process monitoring and diagnosis

In industrial process monitoring, uncertainty in a system arises when measured data are not representative of actual data. Uncertain information should be extracted to maintain safe manufacturing. Interval value methods are generally used to monitor uncertainty in systems, and these methods assume the same uncertainty among variables. However, the variables of the same process differ in uncertainty. Moreover, the dynamic nature of the system causes a more complex uncertainty. To monitor and diagnose such processes, a fault-sensitive bidirectional generative adversarial network (FSBiGAN) is proposed. First, to extract valuable characteristics from uncertain input, BiGANs with self-attentive mechanisms are put in place. Second, a sensitive variable selection approach is utilized to create new fault-sensitive indicators capable of adapting to the dynamic changes of each process. Third, the variable selection approach is applied once more to identify variables causing a fault to occur. Finally, the self-attentive and sensitive variable-picking algorithms can be combined with other generative models to identify defects and analyze the dynamic properties of the process. Comparative experiments applying traditional monitoring and interval methods on the Tennessee Eastman dataset injected with differential uncertainty validate the effectiveness of FSBiGAN in monitoring and diagnosing dynamic uncertainty systems.

In situ health monitoring of multiscale structures and its instantaneous verification using mechanoluminescence and dual machine learning

Extensive changes in the legal, commercial and technical requirements in engineering fields have necessitated automated real-time structural health monitoring (SHM) and instantaneous verification. An integrated system with mechanoluminescence (ML) and dual artificial intelligence (AI) modules with subsidiary finite element method (FEM) simulation is designed for in situ SHM and instantaneous verification. The ML module detects the exact position of a crack tip and evaluates the significance of existing cracks with a plastic stress-intensity factor (PSIF; ). ML fields and their corresponding values are referenced and verified using the FEM simulation and bidirectional generative adversarial network (GAN). Well-trained forward and backward GANs create fake FEM and ML images that appear authentic to observers; a convolutional neural network is used to postulate precise PSIFs from fake images. Finally, the reliability of the proposed system to satisfy existing commercial requirements is validated in terms of tension, compact tension, AI, and instrumentation.

Generative Adversarial Networks for anomaly detection in aerial images

Generative Adversarial Networks (GANs) are commonly used as a system able to perform unsupervised learning. We propose and demonstrate the use of a GAN architecture, known as the fast Anomaly Generative Adversarial Network (f-AnoGAN), to solve the problem of anomaly detection from aerial images. This architecture was previously applied to medical images and, in this work, we adapt it for use on satellite or aerial photographs. To test the effectiveness of this approach, we implemented anomaly detection schemes based on the Bi-directional Generative Adversarial Network (BiGAN), the image-z-image mapping (izi), the z-image-z (ziz) mapping, and a deep convolutional autoencoder (AE). The results show that the f-AnoGAN outperformed others, achieving AUC (area under the curve) values of 0.99 and 0.92 for urban and rural spaces image sets, respectively.

Online Power System Event Detection via Bidirectional Generative Adversarial Networks

Accurate and speedy detection of power system events is critical to enhancing the reliability and resiliency of power systems. Although supervised deep learning algorithms show great promise in identifying power system events, they require a large volume of high-quality event labels for training. This paper develops a bidirectional anomaly generative adversarial network (GAN)-based algorithm to detect power system events using streaming PMU data, which does not rely on a huge amount of event labels. By introducing conditional entropy constraint in the objective function of GAN and graph signal processing-based PMU sorting technique, our proposed algorithm significantly outperforms state-of-the-art event detection algorithms in terms of accuracy. To facilitate the adoption of the proposed algorithm, a prototype online platform is also developed using Apache Hadoop, Kafka, and Spark to enable real-time event detection. The accuracy and computational efficiency of the proposed algorithm are validated using a large-scale real-world PMU dataset from the Eastern Interconnection of the United States.

Forecasting crude oil risk: A multiscale bidirectional generative adversarial network based approach

The crude oil market is known to be subject to the influence of transient and extreme events. The rare and infrequent nature of these events leads to problems such as a lack of data for the estimation of reliable risk measures in the crude oil market. In this paper, an innovative MEMD-BiGAN risk forecasting methodology combining the power of multi scale analysis and Generative Adversarial Network has been proposed. BiGAN has been introduced as an innovative method from the machine learning field to produce the augmented dataset with sufficient number of observations. Then Historical Simulation method can be employed to estimate market risk level in the multiscale domain, where the final risk forecasts taking into account the transient risk factors are more accurate and reliable. MEMD-BiGAN has been applied to model the portfolio of daily trading data in the major crude oil markets including West Texas Intermediate, Brent and OPEC markets. Results suggest that the MEMD-BiGAN model can achieve improved risk coverage. This implies that the incorporation of the transient risk factors using the BiGAN model is essential to more accurate modeling of the risk measures in the turbulent market environment.

How much BiGAN and CycleGAN-learned hidden features are effective for COVID-19 detection from CT images? A comparative study

Bidirectional generative adversarial networks (BiGANs) and cycle generative adversarial networks (CycleGANs) are two emerging machine learning models that, up to now, have been used as generative models, i.e., to generate output data sampled from a target probability distribution. However, these models are also equipped with encoding modules, which, after weakly supervised training, could be, in principle, exploited for the extraction of hidden features from the input data. At the present time, how these extracted features could be effectively exploited for classification tasks is still an unexplored field. Hence, motivated by this consideration, in this paper, we develop and numerically test the performance of a novel inference engine that relies on the exploitation of BiGAN and CycleGAN-learned hidden features for the detection of COVID-19 disease from other lung diseases in computer tomography (CT) scans. In this respect, the main contributions of the paper are twofold. First, we develop a kernel density estimation (KDE)-based inference method, which, in the training phase, leverages the hidden features extracted by BiGANs and CycleGANs for estimating the (a priori unknown) probability density function (PDF) of the CT scans of COVID-19 patients and, then, in the inference phase, uses it as a target COVID-PDF for the detection of COVID diseases. As a second major contribution, we numerically evaluate and compare the classification accuracies of the implemented BiGAN and CycleGAN models against the ones of some state-of-the-art methods, which rely on the unsupervised training of convolutional autoencoders (CAEs) for attaining feature extraction. The performance comparisons are carried out by considering a spectrum of different training loss functions and distance metrics. The obtained classification accuracies of the proposed CycleGAN-based (resp., BiGAN-based) models outperform the corresponding ones of the considered benchmark CAE-based models of about 16% (resp., 14%).

A One-Class Generative Adversarial Detection Framework for Multifunctional Fault Diagnoses

In this article, fault diagnosis is of great significance for system health maintenance. For real applications, diagnosis accuracy suffers from unbalanced data patterns, where normal data are usually abundant than anomaly ones, leading to tremendous diagnosis obstacles. Therefore, it is challenging to use only normal data for fault diagnosis under this imbalanced condition. In addition, a single fault diagnosis model can only conduct one fault diagnosis task in most of cases. Accordingly, a one-class generative adversarial detection (OCGAD) framework based on semisupervised learning is proposed to learn one-class latent knowledge for dealing with multiple semisupervised fault diagnosis tasks, i.e., fault detection using only normal knowledge learning, novelty detection from unknown conditional data, and fault classification with unlabeled data. A bi-directional generative adversarial network (Bi-GAN) is first trained with only normal data. A one-class support vector machine is then established using features exacted by Bi-GAN from signals acquired from an attitude sensor for multifunctional fault detection. The presented OCGAD model is validated using an industrial robot with experiments of three fault detection tasks. The results demonstrate that the present model has good performance for dealing with multiple semisupervised diagnosis problems.

Application of Bidirectional Generative Adversarial Networks to Predict Potential miRNAs Associated With Diseases.

Substantial evidence has shown that microRNAs are crucial for biological processes within complex human diseases. Identifying the association of miRNA–disease pairs will contribute to accelerating the discovery of potential biomarkers and pathogenesis. Researchers began to focus on constructing computational models to facilitate the progress of disease pathology and clinical medicine by identifying the potential disease-related miRNAs. However, most existing computational methods are expensive, and their use is limited to unobserved relationships for unknown miRNAs (diseases) without association information. In this manuscript, we proposed a creatively semi-supervised model named bidirectional generative adversarial network for miRNA-disease association prediction (BGANMDA). First, we constructed a microRNA similarity network, a disease similarity network, and Gaussian interaction profile kernel similarity based on the known miRNA–disease association and comprehensive similarity of miRNAs (diseases). Next, an integrated similarity feature network with the full underlying relationships of miRNA–disease pairwise was obtained. Then, the similarity feature network was fed into the BGANMDA model to learn advanced traits in latent space. Finally, we ranked an association score list and predicted the associations between miRNA and disease. In our experiment, a five-fold cross validation was applied to estimate BGANMDA’s performance, and an area under the curve (AUC) of 0.9319 and a standard deviation of 0.00021 were obtained. At the same time, in the global and local leave-one-out cross validation (LOOCV), the AUC value and standard deviation of BGANMDA were 0.9116 ± 0.0025 and 0.8928 ± 0.0022, respectively. Furthermore, BGANMDA was employed in three different case studies to validate its prediction capability and accuracy. The experimental results of the case studies showed that 46, 46, and 48 of the top 50 prediction lists had been identified in previous studies.

Improved Bidirectional GAN-Based Approach for Network Intrusion Detection Using One-Class Classifier

Existing generative adversarial networks (GANs), primarily used for creating fake image samples from natural images, demand a strong dependence (i.e., the training strategy of the generators and the discriminators require to be in sync) for the generators to produce as realistic fake samples that can “fool” the discriminators. We argue that this strong dependency required for GAN training on images does not necessarily work for GAN models for network intrusion detection tasks. This is because the network intrusion inputs have a simpler feature structure such as relatively low-dimension, discrete feature values, and smaller input size compared to the existing GAN-based anomaly detection tasks proposed on images. To address this issue, we propose a new Bidirectional GAN (Bi-GAN) model that is better equipped for network intrusion detection with reduced overheads involved in excessive training. In our proposed method, the training iteration of the generator (and accordingly the encoder) is increased separate from the training of the discriminator until it satisfies the condition associated with the cross-entropy loss. Our empirical results show that this proposed training strategy greatly improves the performance of both the generator and the discriminator even in the presence of imbalanced classes. In addition, our model offers a new construct of a one-class classifier using the trained encoder–discriminator. The one-class classifier detects anomalous network traffic based on binary classification results instead of calculating expensive and complex anomaly scores (or thresholds). Our experimental result illustrates that our proposed method is highly effective to be used in network intrusion detection tasks and outperforms other similar generative methods on two datasets: NSL-KDD and CIC-DDoS2019 datasets.

{{mathrm {Latent}}Out}: an unsupervised deep anomaly detection approach exploiting latent space distribution

Anomaly detection methods exploiting autoencoders (AE) have shown good performances. Unfortunately, deep non-linear architectures are able to perform high dimensionality reduction while keeping reconstruction error low, thus worsening outlier detecting performances of AEs. To alleviate the above problem, recently some authors have proposed to exploit Variational autoencoders (VAE) and bidirectional Generative Adversarial Networks (GAN), which arise as a variant of standard AEs designed for generative purposes, both enforcing the organization of the latent space guaranteeing continuity. However, these architectures share with standard AEs the problem that they generalize so well that they can also well reconstruct anomalies. In this work we argue that the approach of selecting the worst reconstructed examples as anomalies is too simplistic if a continuous latent space autoencoder-based architecture is employed. We show that outliers tend to lie in the sparsest regions of the combined latent/error space and propose the mathrm{VAE}Out and {{mathrm {Latent}}Out} unsupervised anomaly detection algorithms, identifying outliers by performing density estimation in this augmented feature space. The proposed approach shows sensible improvements in terms of detection performances over the standard approach based on the reconstruction error.