Domain Adversarial Network Research Articles

Multiscale convolutional conditional domain adversarial network with channel attention for unsupervised bearing fault diagnosis

Unsupervised cross-domain fault diagnosis is an effective technical way to realize the engineering application of bearing fault diagnosis methods. However, there are still two problems that need to be resolved. First, the importance of fault features at different scales is generally not consistent. There is redundant information in the fault features. Second, most methods mainly study how to lessen the marginal distribution difference in source and target domains while ignoring their class information. When the data distribution contains complex multimodal structure, this may lead to failure to capture the multimodal structure. To address the above problems, a multiscale channel attention conditional domain adversarial network is proposed. First, a new channel attention module is designed to assign different weights to different channels, which can highlight valuable features and stamp out superfluous features. Then, conditional domain adversarial is used to fully capture the multimodal structure through cross-covariance dependencies between features and classes. Our method’s capability is validated by diagnose results on public data sets and self-built data sets.

Across Sessions and Subjects Domain Adaptation for Building Robust Myoelectric Interface.

Gesture interaction via surface electromyography (sEMG) signal is a promising approach for advanced human-computer interaction systems. However, improving the performance of the myoelectric interface is challenging due to the domain shift caused by the signal's inherent variability. To enhance the interface's robustness, we propose a novel adaptive information fusion neural network (AIFNN) framework, which could effectively reduce the effects of multiple scenarios. Specifically, domain adversarial training is established to inhibit the shared network's weights from exploiting domain-specific representation, thus allowing for the extraction of domain-invariant features. Effectively, classification loss, domain diversence loss and domain discrimination loss are employed, which improve classification performance while reduce distribution mismatches between the two domains. To simulate the application of myoelectric interface, experiments were carried out involving three scenarios (intra-session, inter-session and inter-subject scenarios). Ten non-disabled subjects were recruited to perform sixteen gestures for ten consecutive days. The experimental results indicated that the performance of AIFNN was better than two other state-of-the-art transfer learning approaches, namely fine-tuning (FT) and domain adversarial network (DANN). This study demonstrates the capability of AIFNN to maintain robustness over time and generalize across users in practical myoelectric interface implementations. These findings could serve as a foundation for future deployments.

Adversarial subdomain adaptation method based on multi-scale features for bearing fault diagnosis

Due to the variable working environment of bearings, the collected data often follow different probability distributions. It is hard to directly use the trained models to identify the bearing fault with different operating conditions. In addition, it is a high cost to label the samples for every work condition. To solve these problems, a multi-scale adversarial subdomain adaptation bearing fault diagnosis method is proposed, which is based on Continuous Wavelet Transform (CWT) and our constructed Multi-scale Adversarial SubDomain Adaptation Network (MASDAN). Firstly, to extract the features of non-stationary signals with different frequency bands, CWT is used to convert continuous vibration signals into two-dimensional time-frequency images. Secondly, to enhance the correlation of the features across frequency bands, a multi-scale ConvNeXt is proposed, which adds a multi-scale module based on ConvNeXt to obtain features with different scales. Finally, to reduce the distribution discrepancy between the source domain and the target domain and to avoid feature confusion in different domains, a domain adaptive alignment network introducing domain information is constructed. Two modules are included: the Domain Alignment Classification Network Module (DACNM) based on Multi-kernel Local Maximum Mean Discrepancy (MK-LMMD) and the Domain Adversarial Network Module (DANM) based on domain discrimination. Thus, the MASDAN consists of the multi-scale ConvNeXt module, DACNM, and DANM, which can realize the multi-scale feature extraction and adaptively align the fault features at different domains. The experimental results on the Qingdao University of Technology (QUT) bearing dataset and the Case Western Reserve University (CWRU) bearing dataset demonstrate the proposed method can effectively diagnose bearing faults under different operating conditions.

Cross-Conditions Fault Diagnosis of Rolling Bearing Based on Transitional Domain Adversarial Network

Cross-Conditions Fault Diagnosis of Rolling Bearing Based on Transitional Domain Adversarial Network

A dual-weight domain adversarial network for partial domain fault diagnosis of feedwater heater system

Domain adaptation (DA) approaches have received significant attention in industrial cross-domain fault diagnosis. However, the scarcity of sufficient labeled fault data, particularly under varying loading conditions and harsh operational environments, can give rise to distinct label spaces between two domains, thereby impeding the application of DA-based diagnosis methods. In this paper, we propose a novel dual-weight domain adversarial network (DWDAN) for diagnosing partial domain faults of feedwater heater system in a large-scale power unit, where the target label space is a subset of the source domain. Firstly, domain adversarial network with an instance-based feature learning strategy is constructed to capture domain-invariant and class-discriminative features hidden in raw process data, thereby enhancing feature extraction and generalization abilities of fault diagnosis. Furthermore, a dual-stage reweighted induction module is designed to quantify the contribution of samples from both class-level and sample-level for selective adaptation. This module can automatically eliminate outlier fault categories in the source domain and facilitates alignment of feature distributions for shared fault categories. Comprehensive experiments conducted on the feedwater heater system of a 600 MW coal-fired generating unit demonstrate the outstanding performance of DWDAN.

DCAMIL: Eye-tracking guided dual-cross-attention multi-instance learning for refining fundus disease detection

Deep neural networks (DNNs) have facilitated the development of computer-aided diagnosis (CAD) systems for fundus diseases, helping ophthalmologists to reduce missed diagnoses and misdiagnosis rates. However, the majority of CAD systems are data-driven, but lack the prior medical knowledge that can be performance-friendly. In this regard, we innovatively proposed a human-in-the-loop (HITL) CAD system by leveraging ophthalmologists’ eye-tracking information. Concretely, the HITL CAD system was implemented on the multi-instance learning (MIL), where clinicians’ gaze maps were beneficial to cherry-pick diagnosis-related instances. Furthermore, the dual-cross-attention MIL (DCAMIL) network was utilized to curb the adverse effects of noisy instances. Meanwhile, both the sequence augmentation (SA) module and the domain adversarial network (DAN) were introduced to enrich and standardize the instances in the training bag, respectively, thereby enhancing the robustness of our method. We conduct comparative experiments on our newly-constructed datasets (namely, AMD-Gaze and DR-Gaze) for the AMD and early DR detection, respectively. Rigorous experiments demonstrate the feasibility of our HITL CAD system and the superiority of the proposed DCAMIL, which fully exploits ophthalmologists’ eye-tracking information. These investigations indicate that clinicians’ gaze maps, as prior medical knowledge, is potential to contribute to the CAD systems of clinical diseases.

Cross-condition quantitative diagnosis method for bearing faults based on IDRSN-ECDAN

The development of cross-condition quantitative diagnosis methods for estimating bearing faults based on deep transfer learning technology is considered of great importance for practical applications. However, the existing intelligent quantitative diagnostic methods suffer from noise interference in the vibration data and require the utilization of historical data. Therefore, in this work, a cross-condition quantitative diagnostic method for estimating the bearing faults based on an improved deep residual shrinkage network—entropy conditional domain adversarial network (IDRSN-ECDAN) was proposed. First, a sub-network was added to the residual module to construct a residual shrinkage module, which reduced the noise interference in vibration signals. Next, DropBlock layers were added to the deep residual shrinkage network, and the Adamax optimizer was adopted to improve the diagnostic ability of the model further. Finally, the IDRSN was combined with the ECDAN to transfer the effective information from the source domain data to the target domain through adversarial training. The proposed method was used to systematically analyse a bearing dataset with 15 different fault sizes from a doubly-fed wind turbine test platform, and its superiority was demonstrated through performing several experiments on cross-condition bearing fault quantitative diagnosis.

Towards a Corpus (and Language)-Independent Screening of Parkinson's Disease from Voice and Speech through Domain Adaptation.

End-to-end deep learning models have shown promising results for the automatic screening of Parkinson's disease by voice and speech. However, these models often suffer degradation in their performance when applied to scenarios involving multiple corpora. In addition, they also show corpus-dependent clusterings. These facts indicate a lack of generalisation or the presence of certain shortcuts in the decision, and also suggest the need for developing new corpus-independent models. In this respect, this work explores the use of domain adversarial training as a viable strategy to develop models that retain their discriminative capacity to detect Parkinson's disease across diverse datasets. The paper presents three deep learning architectures and their domain adversarial counterparts. The models were evaluated with sustained vowels and diadochokinetic recordings extracted from four corpora with different demographics, dialects or languages, and recording conditions. The results showed that the space distribution of the embedding features extracted by the domain adversarial networks exhibits a higher intra-class cohesion. This behaviour is supported by a decrease in the variability and inter-domain divergence computed within each class. The findings suggest that domain adversarial networks are able to learn the common characteristics present in Parkinsonian voice and speech, which are supposed to be corpus, and consequently, language independent. Overall, this effort provides evidence that domain adaptation techniques refine the existing end-to-end deep learning approaches for Parkinson's disease detection from voice and speech, achieving more generalizable models.

TDAN: Transferable Domain Adversarial Network for Link Prediction in Heterogeneous Social Networks

Link prediction has received increased attention in social network analysis. One of the unique challenges in heterogeneous social networks is link prediction in new link types without verified link information, such as recommending products to new overseas groups. Existing link prediction models tend to learn type-specific knowledge on specific link types and predict missing or future links on the same link types. However, because of the uncertainty of new link types in the evolving process of social networks, it is difficult to collect sufficient verified link information in new link types. Therefore, we propose the Transferable Domain Adversarial Network ( TDAN ) based on transfer learning to handle the challenge. TDAN exploits transferable type-shared knowledge in historical link types to help predict the unobserved links in new link types. TDAN mainly comprises a structural encoder, a domain discriminator, and an optimization decoder. The structural encoder learns the link representations in a heterogeneous social network. Subsequently, to learn transferable type-shared knowledge, the domain discriminator distinguishes link representations into different link types while minimizing the differences between type-specific knowledge in adversarial training. Inspired by the denoising auto-encoder, the optimization decoder reconstructs the learned type-shared knowledge to eliminate the noise generated during the adversarial training. Extensive experiments on Facebook and YouTube show that TDAN can outperform the state-of-the-art models.

Domain adaptive networks with limited data for rotating machinery fault diagnosis: a case of study of gears

Rotating machinery is one of the most common components in the industry. Therefore, timely and accurate fault diagnosis of rotating machinery is essential for the regular operation of equipment. At present, some achievements have been made in rotating machinery fault diagnosis based on a large number of marked fault data. However, most of the machines are in a normal state in actuality. Especially, the machines run under different loads, so it is costly to collect a large number of labeled fault data under different load distributions. To solve rotating machinery fault diagnosis in different load conditions with limited samples, a domain adaptive group convolutional neural network is proposed. Firstly, the least squares generative adversarial networks were used to expand the limited target sample data. By changing the objective function, the two defects of the low quality of the vibration signal generated by the traditional generative adversarial networks and the unstable training process are optimized. Secondly, the raw vibration signals in the source domain are pre-trained by the group convolutional neural network, and the group training network effectively reduces network parameters. Finally, the source domain signals and target domain signals were trained in domain adversarial networks to diagnose different distributed data in target domains. The proposed method is validated by collecting the raw vibration signals of gears under different loads and different health states, and the effectiveness of the proposed method is proved. Experimental validation shows that the method proposed in this paper achieves an average accuracy improvement of more than 12% compared to other existing methods.

Bearing cross-domain fault diagnosis based on domain adversarial network

Bearing cross-domain fault diagnosis based on domain adversarial network

An attention mechanism-guided domain adversarial network for gearbox fault diagnosis under different operating conditions

In recent years, transfer learning has been widely used in mechanical fault diagnosis with some achievements. However, most transfer learning methods do not perform well in diagnosis when the speed and load change simultaneously. Inspired by the adversarial learning mechanism, a transfer learning method named attention mechanism-guided domain adversarial network (AMDAN) is proposed in this paper. AMDAN regards the convolutional neural networks (CNNs) as the generator of the domain adversarial network to learn mutually invariant features and the domain classifier as the discriminator of the domain adversarial network. Attention mechanism is introduced to take into account the interchannel and intraspace feature fusion to improve the training efficiency. Then, multi-kernel maximum mean discrepancy (MK-MMD) is used to measure the distance of different feature spaces to achieve domain alignment. Finally, the superiority of AMDAN is verified by two sets of gear fault diagnosis experiments. The experimental results show that AMDAN has the highest classification accuracy and the strongest generalization ability compared with other methods.

Fault Diagnosis of Marine Diesel Engines under Partial Set and Cross Working Conditions Based on Transfer Learning

The development of intelligent ships has an urgent demand for intelligent fault diagnosis technology. The working conditions and fault modes of high-power marine diesel engines gradually tend to be diversified and complicated, and the problems of reliability and safety are becoming more and more prominent. There are a lot of working condition data that lack fault labels, and the fault modes are asymmetric among different working conditions, so it is urgent to study effective fault diagnosis methods. Taking a marine diesel engine as the case validation object, we set up cross-condition and partial set fault diagnosis scenarios, proposed transferring knowledge from the source condition to the target condition for the problem of the lack of fault labels in the target condition, and designed a multi-scale and multi-view domain adversarial network (MMDAN) method for experimental validation using 6S50MC-C7 marine diesel engine system operation data. According to the experimental results, the average diagnostic accuracy of this method reached 96.58%, with a short processing time. Furthermore, it exhibits superior diagnostic performance compared to other transfer learning models in the cross-condition partial set transfer task. Additionally, the method proposed in this paper also offers a new approach and reference for the intelligent diagnosis of other equipment in ships.

Fault Diagnosis in Wind Turbines Based on Weighted Joint Domain Adversarial Network Under Various Working Conditions

In recent years, the practical cross-domain wind turbine fault diagnosis is constrained by label space and distribution differences of fault data under different working conditions. Due to the difference in feature distribution, the generalizability of traditional transfer learning models can be affected. Moreover, the difference in label space, which refers to the target domain (SCADA data of another wind turbine), covers only a subset of the source domain (SCADA data of one wind turbine) fault categories, it reduces the accuracy of recognition by the diagnostic model for the target domain failure. Therefore, a weighted joint domain adversarial network (WJDAN) is proposed in this paper to overcome the above problems. In this method, a weighting function is introduced to identify and remove irrelevant source domain fault categories, with the contribution of source domain data to the domain discriminator, maximum mean discrepancy (MMD) and classifier measured. Besides, the domain adversarial network with the intraclass MMD loss is introduced to minimize the marginal and conditional distributional discrepancies between the source domain and target domain simultaneously. As revealed by the experiments on SCADA data of different wind turbines, the proposed WJDAN outperforms other traditional transfer learning methods on wind turbine fault diagnosis.

Deep Domain Adaptation Enhances Amplification Curve Analysis for Single-Channel Multiplexing in Real-Time PCR.

Data-driven approaches for molecular diagnostics are emerging as an alternative to perform an accurate and inexpensive multi-pathogen detection. A novel technique called Amplification Curve Analysis (ACA) has been recently developed by coupling machine learning and real-time Polymerase Chain Reaction (qPCR) to enable the simultaneous detection of multiple targets in a single reaction well. However, target classification purely relying on the amplification curve shapes currently faces several challenges, such as distribution discrepancies between different data sources of synthetic DNA and clinical samples (i.e., training vs testing). Optimisation of computational models is required to achieve higher performance of ACA classification in multiplex qPCR through the reduction of those discrepancies. Here, we proposed a novel transformer-based conditional domain adversarial network (T-CDAN) to eliminate data distribution differences between the source domain (synthetic DNA data) and the target domain (clinical isolate data). The labelled training data from the source domain and unlabelled testing data from the target domain are fed into the T-CDAN, which learns both domains' information simultaneously. After mapping the inputs into a domain-irrelevant space, T-CDAN removes the feature distribution differences and provides a clearer decision boundary for the classifier, resulting in a more accurate pathogen identification. Evaluation of 198 clinical isolates containing three types of carbapenem-resistant genes (blaNDM, blaIMP and blaOXA-48) illustrates a curve-level accuracy of 93.1% and a sample-level accuracy of 97.0% using T-CDAN, showing an accuracy improvement of 20.9% and 4.9% respectively, compared with previous methods. This research emphasises the importance of deep domain adaptation to enable high-level multiplexing in a single qPCR reaction, providing a solid approach to extend qPCR instruments' capabilities without hardware modification in real-world clinical applications.

Unsupervised domain adaptation methods for cross-species transfer of regulatory code signals.

Due to advances in NGS technologies whole-genome maps of various functional genomic elements were generated for a dozen of species, however experiments are still expensive and are not available for many species of interest. Deep learning methods became the state-of-the-art computational methods to analyze the available data, but the focus is often only on the species studied. Here we take advantage of the progresses in Transfer Learning in the area of Unsupervised Domain Adaption (UDA) and tested nine UDA methods for prediction of regulatory code signals for genomes of other species. We tested each deep learning implementation by training the model on experimental data from one species, then refined the model using the genome sequence of the target species for which we wanted to make predictions. Among nine tested domain adaptation architectures non-adversarial methods Minimum Class Confusion (MCC) and Deep Adaptation Network (DAN) significantly outperformed others. Conditional Domain Adversarial Network (CDAN) appeared as the third best architecture. Here we provide an empirical assessment of each approach using real world data. The different approaches were tested on ChIP-seq data for transcription factor binding sites and histone marks on human and mouse genomes, but is generalizable to any cross-species transfer of interest. We tested the efficiency of each method using species where experimental data was available for both. The results allows us to assess how well each implementation will work for species for which only limited experimental data is available and will inform the design of future experiments in these understudied organisms. Overall, our results proved the validity of UDA methods for generation of missing experimental data for histone marks and transcription factor binding sites in various genomes and highlights how robust the various approaches are to data that is incomplete, noisy and susceptible to analytic bias.

A comprehensive gas recognition algorithm with label-free drift compensation based on domain adversarial network

For most applications, the sensor system needs to identify/measure target gases in mixture conditions simultaneously, while countering the performance degradation caused by the drift deviation and batch to batch discrepancy. To solve these problems, we propose a comprehensive algorithm, hybrid attention-based transformer network with domain adversarial learning (HATN-DA). Inside it, multi-head transformer and channel-wise attention mechanism serving as main blocks of the feature extractor effectively capture temporal dependencies and lays emphasis on the diversities of sensors. Then, the capsule vector is applied in the predictor to encode gas species into the concentration information. Besides, the domain discriminator calculates Wasserstein distance of gas features between source and target domains, which is optimized through the back propagation by the feature extractor during the training process. Based on this, label-free drift compensation and transfer calibration effects are achieved: for public dataset, HATN-DA achieves 97.50∼100% accuracies for drift compensation experiment; for dataset of mixture gases, in the batch transfer tasks, HATN-DA achieves an accuracy of 98.79%, significantly improved compared to 90.16% (before transfer). Besides, in the cross-domain real-time concentration prediction tasks, HATN-DA achieves a normalized root mean squared error of 3.32%, 3.25%, 6.62% and 3.01% for ethanol, methane, carbon monoxide and ethylene.

Modulation Classifier: A Few-Shot Learning Semi-Supervised Method Based on Multimodal Information and Domain Adversarial Network

This letter addresses the issue of underfitting or failure of deep learning models caused by insufficient training samples. Unlike previous supervised methods, a new few-shot learning semi-supervised automatic modulation recognition method based on multimodal information and domain adversarial network is proposed herein. The fusion input of multimodal information realizes the joint utilization of modulated signal modal features in the time and frequency domains. Domain adversarial training mines the potential knowledge information of a large number of unlabeled target domain data and introduces the convolutional block attention module (CBAM) to enhance the ability of the network to represent the key features of data. Numerical results validate the high-average classification accuracy of the proposed scheme compared to that of state-of-the-art schemes using fewer samples, particularly in high-order modulation classification. Alternative network structures are compared to confirm the applicability of multimodal information, domain adversarial training, and CBAM.

A New Probability Guided Domain Adversarial Network for Bearing Fault Diagnosis

Recently, unsupervised domain adaptation (UDA) is widely used in fault diagnosis. Many UDA methods reduce the domain differences through alignment tasks, which allows cross-domain diagnosis of the target domain. However, the alignment and classification tasks are independent in these methods. The alignment task does not actively serve the classification task. Accordingly, the semantic features are polluted in the alignment task, and the prediction certainty of the target domain samples is insufficient. In this study, a new probability guided domain adversarial network (PG-DAN) is proposed. A new domain alignment loss is designed. Specifically, classification ability is used as a measure of domain differences to avoid the contamination of semantic features by alignment tasks. In particular, a classifier is used as a discriminator and the classifier prediction information for feature alignment. Although the prediction with high-confidence probability can be obtained through the abovementioned module, misclassification is still possible. To avoid the misclassification problem, a gradient supervision module is designed to guide alignment tasks. This module explicitly reduces the gradient difference caused by misclassification. Comprehensive experiments show that our method can achieve better diagnostic performance.

A Novel Domain Adversarial Networks Based on 3D-LSTM and Local Domain Discriminator for Hearing-Impaired Emotion Recognition.

Recent research on emotion recognition suggests that deep network-based adversarial learning has an ability to solve the cross-subject problem of emotion recognition. This study constructed a hearing-impaired electroencephalography (EEG) emotion dataset containing three emotions (positive, neutral, and negative) in 15 subjects. The emotional domain adversarial neural network (EDANN) was carried out to identify hearing-impaired subjects' emotions by learning hidden emotion information between the labeled data and the data with no-label. For the input data, we propose a spatial filter matrix to reduce the overfitting of the training data. A feature extraction network 3DLSTM-ConvNET was used to extract comprehensive emotional information from the time, frequency, and spatial dimensions. Moreover, emotion local domain discriminator and emotion film group local domain discriminator were added to reduce the distribution distance between the same kinds of emotions and different film groups, respectively. According to the experimental results, the average accuracy of subject-dependent is 0.984 (STD: 0.011), and that of subject-independent is 0.679 (STD: 0.140). In addition, by analyzing the discrimination characteristics, we found that the brain regions with emotional recognition in the hearing-impaired are distributed in the wider areas of the parietal and occipital lobes, which may be caused by visual processing.