Adversarial Learning Framework Research Articles

Graph-based domain adversarial learning framework for video anomaly detection domain generalization

Graph-based domain adversarial learning framework for video anomaly detection domain generalization

Anomaly classification based on self-supervised learning and its application

BackgroundAnomaly classification remains a challenging task in computer vision applications across diverse practical fields such as industrial detection and security check. The purpose of this study is to develop a new anomaly classification method based on self-supervised learning (Self-ACM), which is expected to enhance anomaly classification accuracy significantly. MethodsThe feature maps of images were abstracted using VGG16 model pre-trained on ImageNet, which were then fed into the innovative model to capture the normality distribution within the dataset domain. Then, a self-supervised adversarial anomaly classification learning framework was proposed to facilitate the acquisition of a higher-level semantic representations for improved anomaly detection. Thirdly, we collected and constructed a novel terahertz (THZ) dataset, which serves as a pioneering resource for benchmarking anomaly classification tasks in the field. ResultsThrough a series of rigorous experiments, our findings unequivocally demonstrate the following key insights: Firstly, harnessing feature maps as input data yielded a significant enhancement in anomaly detection performance, underscoring the effectiveness of this approach. Secondly, the integration of self-supervision enriched the dataset with invaluable information, empowering both the discriminator and generator to acquire superior feature representations. The culmination of these advancements is our novel method achieving unparalleled state-of-the-art performance across multiple benchmark datasets. This breakthrough underscores the transformative impact of our approach on anomaly detection methodologies, solidifying its position as a pioneering solution in the field. ConclusionThe Self-ACM strategy not only advances anomaly detection methodologies but also offers a remarkable contribution to dataset creation, setting a new standard for anomaly classification research.

An improved method for rapid un-collapsed building extraction from post-disaster high-resolution remote sensing imagery based on multi-scale feature alignment

ABSTRACT Quick and accurate extraction of un-collapsed buildings from post-disaster High-resolution Remote Sensing Images (HRSIs) is imperative for emergency response. Pre-disaster HRSIs could serve as auxiliary data for training models to expedite this extraction process. However, the effectiveness of models trained directly on pre-disaster HRSIs tends to diminish when applied to post-disaster scenarios, mainly due to the notable discrepancies between these datasets. The current popular approach to mitigate this issue involves aligning features from pre- and post-disaster images using an unsupervised domain adversarial learning framework. However, conventional methods often fall short in reducing the substantial disparity between pre- and post-disaster images, due to a lack of comprehensive alignment of category-level and multi-scale features. To overcome these limitations, we propose the Multi-scale Global and Category-attention Features Alignment Network (MGCAN). This novel approach further refines feature alignment strategies by concurrently aligning both multi-scale global and category-attention features, thus effectively narrowing the gap between pre- and post-disaster HRSIs. Extensive experiments have demonstrated that MGCAN significantly improves the accuracy of un-collapsed building extraction from post-disaster HRSIs. Moreover, compared to other state-of-the-art domain adversarial networks, MGCAN exhibits superior performance in different disaster scenarios.

ScGAL: unmask tumor clonal substructure by jointly analyzing independent single-cell copy number and scRNA-seq data

BackgroundAccurately deciphering clonal copy number substructure can provide insights into the evolutionary mechanism of cancer, and clustering single-cell copy number profiles has become an effective means to unmask intra-tumor heterogeneity (ITH). However, copy numbers inferred from single-cell DNA sequencing (scDNA-seq) data are error-prone due to technically confounding factors such as amplification bias and allele-dropout, and this makes it difficult to precisely identify the ITH.ResultsWe introduce a hybrid model called scGAL to infer clonal copy number substructure. It combines an autoencoder with a generative adversarial network to jointly analyze independent single-cell copy number profiles and gene expression data from same cell line. Under an adversarial learning framework, scGAL exploits complementary information from gene expression data to relieve the effects of noise in copy number data, and learns latent representations of scDNA-seq cells for accurate inference of the ITH. Evaluation results on three real cancer datasets suggest scGAL is able to accurately infer clonal architecture and surpasses other similar methods. In addition, assessment of scGAL on various simulated datasets demonstrates its high robustness against the changes of data size and distribution. scGAL can be accessed at: https://github.com/zhyu-lab/scgal.ConclusionsJoint analysis of independent single-cell copy number and gene expression data from a same cell line can effectively exploit complementary information from individual omics, and thus gives more refined indication of clonal copy number substructure.

RobEns: Robust Ensemble Adversarial Machine Learning Framework for Securing IoT Traffic.

Recently, Machine Learning (ML)-based solutions have been widely adopted to tackle the wide range of security challenges that have affected the progress of the Internet of Things (IoT) in various domains. Despite the reported promising results, the ML-based Intrusion Detection System (IDS) proved to be vulnerable to adversarial examples, which pose an increasing threat. In fact, attackers employ Adversarial Machine Learning (AML) to cause severe performance degradation and thereby evade detection systems. This promoted the need for reliable defense strategies to handle performance and ensure secure networks. This work introduces RobEns, a robust ensemble framework that aims at: (i) exploiting state-of-the-art ML-based models alongside ensemble models for IDSs in the IoT network; (ii) investigating the impact of evasion AML attacks against the provided models within a black-box scenario; and (iii) evaluating the robustness of the considered models after deploying relevant defense methods. In particular, four typical AML attacks are considered to investigate six ML-based IDSs using three benchmarking datasets. Moreover, multi-class classification scenarios are designed to assess the performance of each attack type. The experiments indicated a drastic drop in detection accuracy for some attempts. To harden the IDS even further, two defense mechanisms were derived from both data-based and model-based methods. Specifically, these methods relied on feature squeezing as well as adversarial training defense strategies. They yielded promising results, enhanced robustness, and maintained standard accuracy in the presence or absence of adversaries. The obtained results proved the efficiency of the proposed framework in robustifying IDS performance within the IoT context. In particular, the accuracy reached 100% for black-box attack scenarios while preserving the accuracy in the absence of attacks as well.

DanceAnyWay: Synthesizing Beat-Guided 3D Dances with Randomized Temporal Contrastive Learning

We present DanceAnyWay, a generative learning method to synthesize beat-guided dances of 3D human characters synchronized with music. Our method learns to disentangle the dance movements at the beat frames from the dance movements at all the remaining frames by operating at two hierarchical levels. At the coarser "beat" level, it encodes the rhythm, pitch, and melody information of the input music via dedicated feature representations only at the beat frames. It leverages them to synthesize the beat poses of the target dances using a sequence-to-sequence learning framework. At the finer "repletion" level, our method encodes similar rhythm, pitch, and melody information from all the frames of the input music via dedicated feature representations. It generates the full dance sequences by combining the synthesized beat and repletion poses and enforcing plausibility through an adversarial learning framework. Our training paradigm also enforces fine-grained diversity in the synthesized dances through a randomized temporal contrastive loss, which ensures different segments of the dance sequences have different movements and avoids motion freezing or collapsing to repetitive movements. We evaluate the performance of our approach through extensive experiments on the benchmark AIST++ dataset and observe improvements of about 7%-12% in motion quality metrics and 1.5%-4% in motion diversity metrics over the current baselines, respectively. We also conducted a user study to evaluate the visual quality of our synthesized dances. We noted that, on average, the samples generated by our method were about 9-48% more preferred by the participants and had a 4-27% better five-point Likert-scale score over the best available current baseline in terms of motion quality and synchronization. Our source code and project page are available at https://github.com/aneeshbhattacharya/DanceAnyWay.

DiffAIL: Diffusion Adversarial Imitation Learning

Imitation learning aims to solve the problem of defining reward functions in real-world decision-making tasks. The current popular approach is the Adversarial Imitation Learning (AIL) framework, which matches expert state-action occupancy measures to obtain a surrogate reward for forward reinforcement learning. However, the traditional discriminator is a simple binary classifier and doesn't learn an accurate distribution, which may result in failing to identify expert-level state-action pairs induced by the policy interacting with the environment. To address this issue, we propose a method named diffusion adversarial imitation learning (DiffAIL), which introduces the diffusion model into the AIL framework. Specifically, DiffAIL models the state-action pairs as unconditional diffusion models and uses diffusion loss as part of the discriminator's learning objective, which enables the discriminator to capture better expert demonstrations and improve generalization. Experimentally, the results show that our method achieves state-of-the-art performance and significantly surpasses expert demonstration on two benchmark tasks, including the standard state-action setting and state-only settings.

Self-Interested Coalitional Crowdsensing for Multi-Agent Interactive Environment Monitoring.

As a promising paradigm, mobile crowdsensing (MCS) takes advantage of sensing abilities and cooperates with multi-agent reinforcement learning technologies to provide services for users in large sensing areas, such as smart transportation, environment monitoring, etc. In most cases, strategy training for multi-agent reinforcement learning requires substantial interaction with the sensing environment, which results in unaffordable costs. Thus, environment reconstruction via extraction of the causal effect model from past data is an effective way to smoothly accomplish environment monitoring. However, the sensing environment is often so complex that the observable and unobservable data collected are sparse and heterogeneous, affecting the accuracy of the reconstruction. In this paper, we focus on developing a robust multi-agent environment monitoring framework, called self-interested coalitional crowdsensing for multi-agent interactive environment monitoring (SCC-MIE), including environment reconstruction and worker selection. In SCC-MIE, we start from a multi-agent generative adversarial imitation learning framework to introduce a new self-interested coalitional learning strategy, which forges cooperation between a reconstructor and a discriminator to learn the sensing environment together with the hidden confounder while providing interpretability on the results of environment monitoring. Based on this, we utilize the secretary problem to select suitable workers to collect data for accurate environment monitoring in a real-time manner. It is shown that SCC-MIE realizes a significant performance improvement in environment monitoring compared to the existing models.

Dynamic Weighted Adversarial Learning for Semi-Supervised Classification under Intersectional Class Mismatch

Nowadays, class-mismatch problem has drawn intensive attention in Semi-Supervised Learning (SSL), where the classes of labeled data are assumed to be only a subset of the classes of unlabeled data. However, in a more realistic scenario, the labeled data and unlabeled data often share some common classes while they also have their individual classes, which leads to an “intersectional class-mismatch” problem. As a result, existing SSL methods are often confused by these individual classes and suffer from performance degradation. To address this problem, we propose a novel Dynamic Weighted Adversarial Learning (DWAL) framework to properly utilize unlabeled data for boosting the SSL performance. Specifically, to handle the influence of the individual classes in unlabeled data (i.e., Out-Of-Distribution classes), we propose an enhanced adversarial domain adaptation to dynamically assign weight for each unlabeled example from the perspectives of domain adaptation and a class-wise weighting mechanism, which consists of transferability score and prediction confidence value. Besides, to handle the influence of the individual classes in labeled data (i.e., private classes), we propose a dissimilarity maximization strategy to suppress the inaccurate correlations caused by the examples of individual classes within labeled data. Therefore, our DWAL can properly make use of unlabeled data to acquire an accurate SSL classifier under intersectional class-mismatch setting, and extensive experimental results on five public datasets demonstrate the effectiveness of the proposed model over other state-of-the-art SSL methods.

Prompt-Based Learning for Unpaired Image Captioning

Unpaired Image Captioning (UIC) has been developed to learn image descriptions from unaligned vision-language sample pairs. Existing works usually tackle this task using adversarial learning and visual concept reward based on reinforcement learning. However, these existing works were only able to learn limited cross-domain information in vision and language domains, which restrains the captioning performance of UIC. Inspired by the success of Vision-Language Pre-Trained Models (VL-PTMs) in this research, we attempt to infer the cross-domain cue information about a given image from the large VL-PTMs for the UIC task. This research is also motivated by recent successes of prompt learning in many downstream multi-modal tasks, including image-text retrieval and vision question answering. In this work, a semantic prompt is introduced and aggregated with visual features for more accurate caption prediction under the adversarial learning framework. In addition, a metric prompt is designed to select high-quality pseudo image-caption samples obtained from the basic captioning model and refine the model in an iterative manner. Extensive experiments on the COCO and Flickr30 K datasets validate the promising captioning ability of the proposed model. We expect that the proposed prompt-based UIC model will stimulate a new line of research for the VL-PTMs based captioning.

Adversarial Spatiotemporal Contrastive Learning for Electrocardiogram Signals.

Extracting invariant representations in unlabeled electrocardiogram (ECG) signals is a challenge for deep neural networks (DNNs). Contrastive learning is a promising method for unsupervised learning. However, it should improve its robustness to noise and learn the spatiotemporal and semantic representations of categories, just like cardiologists. This article proposes a patient-level adversarial spatiotemporal contrastive learning (ASTCL) framework, which includes ECG augmentations, an adversarial module, and a spatiotemporal contrastive module. Based on the ECG noise attributes, two distinct but effective ECG augmentations, ECG noise enhancement, and ECG noise denoising, are introduced. These methods are beneficial for ASTCL to enhance the robustness of the DNN to noise. This article proposes a self-supervised task to increase the antiperturbation ability. This task is represented as a game between the discriminator and encoder in the adversarial module, which pulls the extracted representations into the shared distribution between the positive pairs to discard the perturbation representations and learn the invariant representations. The spatiotemporal contrastive module combines spatiotemporal prediction and patient discrimination to learn the spatiotemporal and semantic representations of categories. To learn category representations effectively, this article only uses patient-level positive pairs and alternately uses the predictor and the stop-gradient to avoid model collapse. To verify the effectiveness of the proposed method, various groups of experiments are conducted on four ECG benchmark datasets and one clinical dataset compared with the state-of-the-art methods. Experimental results showed that the proposed method outperforms the state-of-the-art methods.

Semantic Distance Adversarial Learning for Text-to-Image Synthesis

Text-to-Image (T2I) synthesis is a cross-modality task that requires a text description as input to generate a realistic and semantically consistent image. To guarantee semantic consis- tency, previous studies regenerate text descriptions from synthetic images and align them with the given descriptions. However, the existing redescription modules lack explicit modeling of their training objectives, which is crucial for reliable measurement of semantic distance between redescriptions and given text inputs. Consequently, the aligned text redescriptions suffer from training bias caused by the emergence of adversarial image samples, unseen semantics, and mistaken contents from low- quality synthesized images. To this end, we propose a SEMantic distance Adversarial learning (SEMA) framework for Text-to- Image synthesis which strengthens semantic consistency from two aspects: 1) We introduce adversarial learning between the image generator and the text redescription module to mutually promote or demote the quality of generated image or text instances. This learning model ensures accurate redescription of image contents, thus diminishing the generation of adversarial image samples. 2) We introduce two-fold semantic distance discrimination (SEM distance) to characterize semantic relevance between matching text or image pairs. The unseen semantics and mistaken contents will be penalized with a large SEM distance. The proposed discrimination method also simplifies the model training process with no need to optimize multiple discriminators. Experimental results on CUB Birds 200 and MS-COCO datasets show that the proposed model outperforms the state-of-the-art methods. Code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/NJUPT-MCC/SEMA</uri> .

CALEB: A Conditional Adversarial Learning Framework to enhance bot detection

The high growth of Online Social Networks (OSNs) over the last few years has allowed automated accounts, known as social bots, to gain ground. As highlighted by other researchers, many of these bots have malicious purposes and tend to mimic human behavior, posing high-level security threats on OSN platforms. Moreover, recent studies have shown that social bots evolve over time by reforming and reinventing unforeseen and sophisticated characteristics, making them capable of evading the current machine learning state-of-the-art bot detection systems. This work is motivated by the critical need to establish adaptive bot detection methods in order to proactively capture unseen evolved bots towards healthier OSNs interactions. In contrast with most earlier supervised ML approaches which are limited by the inability to effectively detect new types of bots, this paper proposes CALEB, a robust end-to-end proactive framework based on the Conditional Generative Adversarial Network (CGAN) and its extension, Auxiliary Classifier GAN (AC-GAN), to simulate bot evolution by creating realistic synthetic instances of different bot types. These simulated evolved bots augment existing bot datasets and therefore enhance the detection of emerging generations of bots before they even appear. Furthermore, we show that our augmentation approach overpasses other earlier augmentation techniques which fail at simulating evolving bots. Extensive experimentation on well established public bot datasets, show that our approach offers a performance boost of up to 10% regarding the detection of new unseen bots. Finally, the use of the AC-GAN Discriminator as a bot detector, has outperformed former ML approaches, showcasing the efficiency of our end to end framework.

AdvMIL: Adversarial multiple instance learning for the survival analysis on whole-slide images

The survival analysis on histological whole-slide images (WSIs) is one of the most important means to estimate patient prognosis. Although many weakly-supervised deep learning models have been developed for gigapixel WSIs, their potential is generally restricted by classical survival analysis rules and fully-supervised learning requirements. As a result, these models provide patients only with a completely-certain point estimation of time-to-event, and they could only learn from the labeled WSI data currently at a small scale. To tackle these problems, we propose a novel adversarial multiple instance learning (AdvMIL) framework. This framework is based on adversarial time-to-event modeling, and integrates the multiple instance learning (MIL) that is much necessary for WSI representation learning. It is a plug-and-play one, so that most existing MIL-based end-to-end methods can be easily upgraded by applying this framework, gaining the improved abilities of survival distribution estimation and semi-supervised learning. Our extensive experiments show that AdvMIL not only could often bring performance improvement to mainstream WSI survival analysis methods at a relatively low computational cost, but also enables these methods to effectively utilize unlabeled data via semi-supervised learning. Moreover, it is observed that AdvMIL could help improving the robustness of models against patch occlusion and two representative image noises. The proposed AdvMIL framework could promote the research of survival analysis in computational pathology with its novel adversarial MIL paradigm.

Image-to-image translation for improved digital holographic reconstruction based on a generative adversarial network learning framework

Based on the generative-discriminative model, a digital holographic reconstruction generative adversarial network is established (DHR-GAN), in which the objective function is optimized on the original model to evaluate the perceptual holographic image. Performing the goals of different optical characteristics, the experimental acquisition system of digital holographic images is designed and constructed for preparing the digital holographic reconstructed image dataset (DHR-dataset), which achieves supervised training and testing of DHR-GAN. From an extensive qualitative and quantitative comparative discussion, the reconstruction results are more stable, with higher contrast and sharper local details, demonstrating the effectiveness of the proposed network model.

Making DeepFakes More Spurious: Evading Deep Face Forgery Detection via Trace Removal Attack

DeepFakes are raising significant social concerns. Although various DeepFake detectors have been developed as forensic countermeasures, these detectors are still vulnerable to attacks. Recently, a few attacks, principally adversarial attacks, have succeeded in cloaking DeepFake images to evade detection. However, these attacks have typical detector-specific designs, which require prior knowledge about the detector, leading to poor transferability. Moreover, these attacks only consider simple security scenarios. Less is known about how effective they are in high-level scenarios where either the detectors or the attacker's knowledge varies. In this paper, we solve the above challenges with presenting a novel detector-agnostic trace removal attack for DeepFake anti-forensics. Instead of investigating the detector side, our attack looks into the original DeepFake creation pipeline, attempting to remove all detectable natural DeepFake traces to render the fake images more "authentic". To implement this attack, first, we perform a DeepFake trace discovery, identifying three discernible traces. Then a trace removal network (TR-Net) is proposed based on an adversarial learning framework involving one generator and multiple discriminators. Each discriminator is responsible for one individual trace representation to avoid cross-trace interference. These discriminators are arranged in parallel, which prompts the generator to remove various traces simultaneously. To evaluate the attack efficacy, we crafted heterogeneous security scenarios where the detectors were embedded with different levels of defense and the attackers' background knowledge of data varies. The experimental results show that the proposed attack can significantly compromise the detection accuracy of six state-of-the-art DeepFake detectors while causing only a negligible loss in visual quality to the original DeepFake samples.

Anchor Link Prediction for Privacy Leakage via De-Anonymization in Multiple Social Networks

Anchor link prediction exacerbates the risk of privacy leakage via the de-anonymization of social network data. Embedding-based methods for anchor link prediction are limited by the excessive similarity of the associated nodes in a latent feature space and the variation between latent feature spaces caused by the semantics of different networks. In this paper, we propose a novel method which reduces the impact of semantic discrepancies between different networks in the latent feature space. The proposed method consists of two phases. First, graph embedding focuses on the network structural roles of nodes and increases the distinction between the associated nodes in the embedding space. Second, a federated adversarial learning framework which performs graph embedding on each social network and an adversarial learning model on the server according to the observable anchor links is used to associate independent graph embedding approaches on different social networks. The combination of distinction enhancement and the association of graph embedding approaches alleviates variance between the latent feature spaces caused by the semantics of different social networks. Extensive experiments on real social networks demonstrate that the proposed method significantly outperforms the state-of-the-art methods in terms of both precision and robustness.

Commonsense-based adversarial learning framework for zero-shot stance detection

Zero-shot stance detection (ZSSD) aims to identify the stance for a diverse range of topics with limited or no training data. It is more suitable for evolving real-world topics that are constantly evolving. However, ZSSD faces major challenges which are the lack of unseen target information and the inability to generalize training information to unseen targets effectively. To overcome these challenges, we propose a commonsense-based adversarial learning framework that comprises a commonsense graph encoder and a feature separation adversarial network. Specifically, our approach utilizes an external commonsense graph encoder to learn unseen target information. Moreover, we design a novel feature separation adversarial network to learn target-invariant and target-specific features, which enhances the model’s ability to reason beyond the seen targets. Experiments on two benchmark datasets show that our proposed framework achieves state-of-the-art performance.

Online adversarial knowledge distillation for graph neural networks

Knowledge distillation, a technique recently gaining popularity for enhancing model generalization in Convolutional Neural Networks (CNNs), operates under the assumption that both teacher and student models are trained on identical data distributions. However, its effect on Graph Neural Networks (GNNs) is less than satisfactory since the graph topology and node attributes are prone to evolve, thereby leading to the issue of distribution shift. In this paper, we tackle this challenge by simultaneously training a group of graph neural networks in an online distillation fashion, where the group knowledge plays a role as a dynamic virtual teacher and the structure changes in graph neural networks are effectively captured. To improve the distillation performance, two types of knowledge are transferred among the students to enhance each other: local knowledge reflecting information in the graph topology and node attributes, and global knowledge reflecting the prediction over classes. We transfer the global knowledge with KL-divergence as the vanilla knowledge distillation does, while exploiting the complicated structure of the local knowledge with an efficient adversarial cyclic learning framework. Extensive experiments verified the effectiveness of our proposed online adversarial distillation approach. The code is published at https://github.com/wangz3066/OnlineDistillGCN.

Transfer easy to hard: Adversarial contrastive feature learning for unsupervised person re-identification

Unsupervised Person Re-Identification (Re-ID) is challenging due to the lack of ground-truth labels. Most existing methods address this problem by progressively mining high-confidence pseudo labels to guide the feature learning process. However, how to construct hard-enough samples while maintaining the fidelity of pseudo labels in these samples remains an open issue in the machine learning community. To tackle this challenge, we design a simple yet effective adversarial contrastive feature learning (ACFL) framework, which enhances the discriminative capability of features by introducing more transformed hard samples in the feature learning process. Specifically, it mainly consists of a discriminative feature learning module and a hard sample generation module. The discriminative feature learning module extracts recognizable features of unlabeled training samples to estimate the high-confidence relationship between samples. Then, the hard sample generation module utilizes these high-confidence relationships between samples to transfer all samples into the hard ones via an adversarial learning strategy. Finally, the generated hard samples are further fed into DFL to learn discriminative features for person Re-ID. Extensive experiments on Market-1501, DukeMTMC-reID, and MSMT17 datasets show that our method compares favorably with state-of-the-art methods.