Adversarial Cross-domain Community Question Retrieval

Recent advances in deep domain adaptation reveal that adversarial learning can be embedded into deep networks to learn transferable features that reduce distribution discrepancy between the source and target domains. Existing domain adversarial adaptation methods based on single domain discriminator only align the source and target data distributions without exploiting the complex multimode structures. In this paper, we present a multi-adversarial domain adaptation (MADA) approach, which captures multimode structures to enable fine-grained alignment of different data distributions based on multiple domain discriminators. The adaptation can be achieved by stochastic gradient descent with the gradients computed by back-propagation in linear-time. Empirical evidence demonstrates that the proposed model outperforms state of the art methods on standard domain adaptation datasets.

Nowadays, the convolutional neural network (CNN) based steganalysis has achieved remarkable performance in the well-controlled lab environment. However, the cover source mismatch (CSM) problem, which can be attributed to the discrepancy between the training, and evaluation datasets, is still one of the pivotal obstacles for adapting the steganalysis into real-world applications. In this letter, we propose to merge the domain adaptation strategy into CNN-based audio steganalysis for handling the CSM problem. Specifically, the proposed framework contains three components: feature extractor, steganalytic classifier, and domain discriminator. The cascade of feature extractor, and steganalytic classifier compose the typical supervised steganalysis model. The unsupervised domain adaptation is implemented by the domain adversarial training between the feature extractor, and domain discriminator. Ultimately, the feature extractor is trained to extract the steganalytic, and domain-invariant features. It aims to reduce the domain gap between the training data, and testing data. The experimental results show that our approach could effectively mitigate the CSM impact caused by the diversity of audio recording devices.

ObjectiveTo examine experiences of racial discrimination among black adults in the United States, which broadly contribute to their poor health outcomes.Data Source and Study DesignData come from a nationally representative, probability‐based telephone survey including 802 non‐Hispanic black and a comparison group of 902 non‐Hispanic white US adults, conducted January–April 2017.MethodsWe calculated the percent of blacks reporting discrimination in several domains, including health care. We used logistic regression to compare the black‐white difference in odds of discrimination, and among blacks only to examine variation by socioeconomic status, gender, and neighborhood racial composition.Principal FindingsAbout one‐third of blacks (32 percent) reported experiencing discrimination in clinical encounters, while 22 percent avoided seeking health care for themselves or family members due to anticipated discrimination. A majority of black adults reported experiencing discrimination in employment (57 percent in obtaining equal pay/promotions; 56 percent in applying for jobs), police interactions (60 percent reported being stopped/unfairly treated by police), and hearing microaggressions (52 percent) and racial slurs (51 percent). In adjusted models, blacks had significantly higher odds than whites of reporting discrimination in every domain. Among blacks, having a college degree was associated with higher odds of experiencing overall institutional discrimination.ConclusionsThe extent of reported discrimination across several areas of life suggests a broad pattern of discrimination against blacks in America, beyond isolated experiences. Black‐white disparities exist on nearly all dimensions of experiences with public and private institutions, including health care and the police. Evidence of systemic discrimination suggests a need for more active institutional interventions to address racism in policy and practice.

Cross-domain aspect-based sentiment analysis aims to utilize the useful knowledge in a source domain to extract aspect terms and predict their sentiment polarities in a target domain. Recently, methods based on adversarial training have been applied to this task and achieved promising results. In such methods, both the source and target data are utilized to learn domain-invariant features through deceiving a domain discriminator. However, the task classifier is only trained on the source data, which causes the aspect and sentiment information lying in the target data can not be exploited by the task classifier. In this paper, we propose an Adaptive Hybrid Framework (AHF) for cross-domain aspect-based sentiment analysis. We integrate pseudo-label based semi-supervised learning and adversarial training in a unified network. Thus the target data can be used not only to align the features via the training of domain discriminator, but also to refine the task classifier. Furthermore, we design an adaptive mean teacher as the semi-supervised part of our network, which can mitigate the effects of noisy pseudo labels generated on the target data. We conduct experiments on four public datasets and the experimental results show that our framework significantly outperforms the state-of-the-art methods.

In order to improve the quality of printed products and promote the application of 3D printing, it is necessary to carry out health monitoring and fault diagnosis for 3D printers. In this paper, an attitude data-based deep transfer capsule network is proposed for intelligent fault diagnosis of delta 3D printers. Based on the forward kinematic analysis, the attitude data change of the moving platform can reflect the fault information of the printers. To extract fault features from the attitude data with rich directional pose information and complete the cross-domain diagnosis task effectively, the proposed approach consists of a feature encoder with capsule layer, a fault pattern classifier, and a domain discriminator. Through the domain adversarial training, the model can minimize the difference between the source domain and the target domain data distribution, and the trained classifier can obtain better diagnosis performance in the target domain. The experiment result demonstrates the superiority and effectiveness of the proposed method for fault diagnosis problems of delta 3D printers.

As the core refrigeration equipment in cooling systems, the operational state of chiller units is crucial for ship support, equipment cooling, and mission stability. However, because of their sensitivity and the complexity of operating environments, obtaining large volumes of complete, fault-labeled data is difficult in practical engineering appli-cations. This limitation makes it challenging for traditional data-driven approaches to deliver accurate fault diagnoses. Furthermore, data collected from different devices or under varying operating conditions often differ significantly in both feature dimensions and distributions, i.e., data heterogeneity, which further complicates model transfer. To address these challenges, this study proposes a deep transfer learning-based fault di-agnosis method designed to leverage abundant knowledge from the source domain while adaptively learning features of the target domain. Given the persistent difficulties in collecting sufficient high-quality labeled fault data, traditional data-driven models continue to face restricted diagnostic performance on target equipment. At the same time, data heterogeneity across devices or operating conditions intensifies the challenge of cross-domain knowledge transfer. To overcome these issues, this study develops a heterogeneous transfer learning method that integrates a dual-channel autoencoder, domain adversarial training, and pseudo-label self-training. This combination enables precise small-sample knowledge transfer from the source to the target domain. Specifi-cally, the dual-channel autoencoder is first applied to align heterogeneous feature di-mensions. Then, a Gradient Reversal Layer (GRL) and a domain discriminator are in-troduced to extract domain-invariant features. In parallel, high-confidence pseu-do-labeled samples from the target domain are incorporated into joint training to im-prove generalization and robustness. Experimental results confirm that the method achieves high fault diagnosis accuracy in typical industrial application scenarios, ena-bling effective identification of common faults in various types of chiller units under conventional operating conditions, the proposed method achieves higher accuracy and F1-scores in multi-class fault diagnosis tasks compared with both traditional approaches and existing transfer learning methods. These findings provide a novel perspective for advancing the intelligent operation and maintenance of chiller units.

In recent years, unsupervised domain adaptation (UDA)-based methods have been widely developed for intelligent bearing fault diagnosis across various working conditions. However, a considerably more challenging and practical fault diagnosis scenario, in which the source and target domains are, respectively, collected from bearings across different positions and machines, is urgent to be addressed. To solve this issue, an innovative end-to-end domain conditioned joint adaptation network (DCJAN), which is composed of a domain conditioned (DC) feature extractor, two classifiers, and a domain discriminator is presented. On the one hand, the DC feature extraction structure is designed to relax totally shared network assumptions in feature extraction and learn more domain-specialized features for cross-domain fault diagnosis of bearings. On the other hand, a joint adaptation strategy is implemented for diagnostic knowledge transfer across domains, in which domain-level and class-level adaptations are, respectively, achieved by domain-adversarial training and bi-classifier adversarial training. Extensive experiments including cross-position fault diagnosis (CPFD) and cross-machine fault diagnosis (CMFD) of bearings indicate the validity and superiority of the proposed method.

For cognitive workload recognition, electroencephalography (EEG) signals vary from different subjects, thus hindering the recognition performance when direct extending to a new subject. Though calibrating the new subject or collecting more data would alleviate this issue, it is generally time-consuming and unrealistic. To cope with the problem, we propose a deep domain adaptation scheme for EEG-based cross-subject cognitive workload recognition, using the knowledge from the existing subjects (source domain) to improve the recognition performance of a new subject (target domain). Specifically, the proposed method has four modules: the EEG features extractor, feature distribution alignment, label classifier, and domain discriminator. The EEG feature extractor learns transferable shallow feature representation of both domains. The label classifier further learns the deep representation from the shallow one and trains the classifier. To reduce the domain discrepancy, we employ feature distribution alignment and domain discriminator from shallow and deep representation views using a distribution discrepancy metric and adversarial training with the feature extractor, respectively. We conduct experiments to recognize the low and high workload levels on a self-designed EEG dataset with 38 subjects performing the working memory cognitive task. Experimental results validate that our proposed framework outperforms the baselines significantly.

Electroencephalography (EEG) is the most prevalent signal acquisition technique for brain-computer interface (BCI). However, the statistical distribution of EEG data varies across subjects and sessions, resulting in poor generalization of the domain-specific classifier. Although the collection of a large number of recordings may alleviate this issue, it is often impractical and not user-friendly. This study proposes the integration of deep domain adaptation with few-shot learning to address the challenge by leveraging the knowledge from multiple source subjects to enhance the performance of a single target subject. The framework incorporated 3 modules: a feature extractor, domain discriminator, and classifier. The feature extractor utilized the available labeled samples with supervised contrastive loss to map the discriminate features onto a deep representation space, where the features from the same class were more similar than those from different classes. The domain discriminator was used to reduce domain drift, through adversarial training. The classifier predicted the user motor intention, based on EEG features. The framework was extensively evaluated through the BCI Competition IV Datasets 2a and 2b. The results of this study indicate that the framework is capable of enhancing the BCI performance and potentially decreases the calibration effort compared to the traditional approach, but the major limitation of this method is that it requires meticulous selection of source subjects.

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.

Class-aware domain adaptation for improving adversarial robustness

By reason of factors such as terrains, weather conditions, sensor imaging methods and cultural and economic development, there is a large shift between the remote sensing imagery collected from different geographic locations and different sensors, which makes the state-of-the-art semantic segmentation models trained on source domain (a image set gathered from specific geographic locations and sensors) difficult to generalize to target domain (another image set collected from other geographic locations and sensors). Currently, unsupervised domain adaptation using adversarial training whose purpose is to align the marginal distribution in the output space between source and target domain, is the most explored and practical approach to address this issue. However, this global alignment approach does not take into account diversities of different regions in a specific image nor the category-level distribution, which leads to the consequence that some regions and categories which are already well aligned between the source and target domain may be incorrectly remapped. Therefore, we propose a region and category adaptive domain discriminator, aiming to emphasize the differences in regions and categories during the process of alignment. Specifically, on the one hand, we propose an entropy-based regional attention module in domain discriminator to emphasize the importance of difficult-to-align regions. On the other hand, we propose a class-clear module to update only the distribution of existing categories in one iteration without affecting all categories. Finally, a lot of experiments are introduced to indicate that the proposed method can obtain better results when compared with other state-of-the-art unsupervised domain adaptation methods using adversarial training.

ASR can be improved by multi-task learning (MTL) with domain enhancing or domain adversarial training, which are two opposite objectives with the aim to increase/decrease domain variance towards domain-aware/agnostic ASR, respectively. In this work, we study how to best apply these two opposite objectives with speaker labels to improve conformer-based ASR. We also propose a novel adaptive gradient reversal layer for stable and effective adversarial training without tuning effort. Detailed analysis and experimental verification are conducted to show the optimal positions in the ASR neural network (NN) to apply speaker enhancing and adversarial training. We also explore their combination for further improvement, achieving the same performance as i-vectors plus adversarial training. Our best speaker-based MTL achieves 7% relative improvement on the Switchboard Hub5'00 set. We also investigate the effect of such speaker-based MTL w.r.t. cleaner dataset and weaker ASR NN.

Domain adversarial training has become a prevailing and effective paradigm for unsupervised domain adaptation (UDA). To successfully align the multi-modal data structures across domains, the following works exploit discriminative information in the adversarial training process, e.g., using multiple class-wise discriminators and involving conditional information in the input or output of the domain discriminator. However, these methods either require non-trivial model designs or are inefficient for UDA tasks. In this work, we attempt to address this dilemma by devising simple and compact conditional domain adversarial training methods. We first revisit the simple concatenation conditioning strategy where features are concatenated with output predictions as the input of the discriminator. We find the concatenation strategy suffers from the weak conditioning strength. We further demonstrate that enlarging the norm of concatenated predictions can effectively energize the conditional domain alignment. Thus we improve concatenation conditioning by normalizing the output predictions to have the same norm of features, and term the derived method as Normalized OutpUt coNditioner (NOUN). However, conditioning on raw output predictions for domain alignment, NOUN suffers from inaccurate predictions of the target domain. To this end, we propose to condition the cross-domain feature alignment in the prototype space rather than in the output space. Combining the novel prototype-based conditioning with NOUN, we term the enhanced method as PROtotype-based Normalized OutpUt coNditioner (PRONOUN). Experiments on both object recognition and semantic segmentation show that NOUN can effectively align the multi-modal structures across domains and even outperform state-of-the-art domain adversarial training methods. Together with prototype-based conditioning, PRONOUN further improves the adaptation performance over NOUN on multiple object recognition benchmarks for UDA. Code is available at https://github.com/tim-learn/NOUN.

Intelligent mechanical fault diagnosis is a crucial measure to ensure the safe operation of equipment. To address the issue of model collapse in domain adversarial training and the problem posed by different training samples having different transferability not considered in transfer tasks, this article proposes a weighted entropy minimization based deep conditional adversarial diagnosis approach of rotating machines under variable working conditions. First, the features of vibration signals in the source domain and target domain are extracted by a weight-sharing one-dimensional deep convolution neural network. The feature vectors and category prediction vectors are then fused by multilinear mapping to carry out adversarial training in domain adaptation. The entropy of the output of the domain discrimination model provides the index by which to measure the transferability of training samples. The transferability weights of samples are applied to the entropy minimization loss to eliminate the influence of these samples that are hard to transfer in adversarial domain adaptation. Experimental datasets under variable working conditions support the value of our approach.