- Research Article
3
- 10.1016/j.neucom.2021.07.033
A novel class restriction loss for unsupervised domain adaptation
- Jul 20, 2021
- Neurocomputing
- Qi He + 3 more +3
A novel class restriction loss for unsupervised domain adaptation
Unsupervised Domain Adaptation (UDA) aims to transfer knowledge from a well-labeled source domain to an unlabeled target domain with a correlative distribution. Numerous existing approaches process this hard nut by directly matching the marginal distribution between two domains, which confront the obstacle of rough alignment and blurred decision boundary. Recent advances in UDA introduce target pseudo-label and subdomain adaptation to reduce misalignment and distribution discrepancy. Whereas, they frequently ignore that the production of target pseudo-label is so dependent on the source-trained classifier, which without reasonable restriction to discriminate generated pseudo-label is whether confident. Meanwhile, many methods in the subdomain alignment metric ignore exploring the potential distribution discrepancy between same-class samples of the intra-domain. To address these two issues simultaneously, this paper proposes a Cycle Consistency based Pseudo Label and Fine Alignment (CCPLFA) approach for UDA. In particular, firstly, a novel cycle-consistency based pseudo label module is designed, which is a simple yet effective way to alleviate the noise of pseudo labels and improve their semantic correctness. Secondly, we develop a Fine-Alignment distribution matching metric. Which can maximize the feature distribution density of intra-class cross-domains and not overlook the distribution structure of the global aspect. Comprehensive experiment results on four benchmarks demonstrate the capability of plug and play and the well generalization performance of our proposed method.
A novel class restriction loss for unsupervised domain adaptation
A novel class restriction loss for unsupervised domain adaptation
Learning Discriminative Geodesic Flow Kernel for Unsupervised Domain Adaptation
Extracting the domain-invariant features provides an important intuition for unsupervised domain adaptation. Due to the unavailable target labels, it is difficult to guarantee that the learned domain-invariant features are good for target instances classification. In this paper, we extend the classic geodesic flow kernel method by leveraging the pseudo labels during the training process to learn a discriminative geodesic flow kernel for unsupervised domain adaptation. Specifically, the proposed method alternately discovers the pseudo target labels and builds the geodesic flow from a discriminative source subspace to another ‘discriminative’ target subspace. More specially, the pseudo target labels are inferred via the learned kernel based on an easy yet effective label propagation strategy. Hence, the proposed method not only holds the property of domain-invariance, but also maximizes the consistency between pseudo label structure and data structure. Experimental results illustrate that the proposed method outperforms the state-of-the-art unsupervised domain adaptation methods for object recognition and sentiment analysis.
Read moreSelf-supervision based Semantic Alignment for Unsupervised Domain Adaptation
Unsupervised domain adaptation aims to learn domain-invariant features across domains to transfer knowledge from a well-labeled source domain to an unlabeled target domain. Recently, some unsupervised domain adaptation methods focus on semantically aligning data distributions with pseudo-labels of the target domain. However, semantic alignment based on pseudo-labels has potential risks, e.g., inaccurate pseudo-labeling from classifier, and error accumulation from pseudo-label bias. To alleviate these risks, we propose a novel self-supervision based semantic alignment (S3A) approach for unsupervised domain adaptation, which can jointly incorporate the source alignment and cross-domain target alignment for better semantic alignment across domains. S3A consists of a two-stage semantic alignment procedure with self-supervision. One is to capture the discriminative structure of source domain by aligning source data to source class prototypes, and the other is to match each target data to its neighbor in source domain with self-supervision. Extensive experiments on three benchmark datasets demonstrate the effectiveness of our proposed method, compared with the representative adversarial learning and self-supervised learning based unsupervised domain adaptation methods.
Read moreSTRUDEL: Self-training with Uncertainty Dependent Label Refinement Across Domains
We propose an unsupervised domain adaptation (UDA) approach for white matter hyperintensity (WMH) segmentation, which uses Self-TRaining with Uncertainty DEpendent Label refinement (STRUDEL). Self-training has recently been introduced as a highly effective method for UDA, which is based on self-generated pseudo labels. However, pseudo labels can be very noisy and therefore deteriorate model performance. We propose to predict the uncertainty of pseudo labels and integrate it in the training process with an uncertainty-guided loss function to highlight labels with high certainty. STRUDEL is further improved by incorporating the segmentation output of an existing method in the pseudo label generation that showed high robustness for WMH segmentation. In our experiments, we evaluate STRUDEL with a standard U-Net and a modified network with a higher receptive field. Our results on WMH segmentation across datasets demonstrate the significant improvement of STRUDEL with respect to standard self-training.
Read moreFeature Re-Representation and Reliable Pseudo Label Retraining for Cross-Domain Semantic Segmentation.
This paper presents a novel unsupervised domain adaptation method for semantic segmentation. We argue that a good representation of the target-domain data should keep both the knowledge from the source domain and the target-domain-specific information. To obtain the knowledge from the source domain, we first learn a set of bases to characterize the feature distribution of the source domain, then features from both the source and the target domain are re-represented as a weighted summation of the source bases. A discriminator is additionally introduced to make the re-representation responsibilities of both domain features under the same bases indistinguishable. In this way, the domain gap between the source re-representation and target re-representation is minimized, and the re-represented target domain features contain the source domain information. Then we combine the feature re-representation with the original domain-specific feature together for subsequent pixel-wise classification. To further make the re-represented target features semantically meaningful, a Reliable Pseudo Label Retraining (RPLR) strategy is proposed, which utilizes the consistency of the prediction by the networks trained with multi-view source images to select the clean pseudo labels on unlabeled target images for re-training. Extensive experiments demonstrate the competitive performance of our approach for unsupervised domain adaptation on the semantic segmentation benchmarks.
Read moreImproving pseudo labels with intra-class similarity for unsupervised domain adaptation
Improving pseudo labels with intra-class similarity for unsupervised domain adaptation
MetaCorrection: Domain-aware Meta Loss Correction for Unsupervised Domain Adaptation in Semantic Segmentation
Unsupervised domain adaptation (UDA) aims to transfer the knowledge from the labeled source domain to the unlabeled target domain. Existing self-training based UDA approaches assign pseudo labels for target data and treat them as ground truth labels to fully leverage unlabeled target data for model adaptation. However, the generated pseudo labels from the model optimized on the source domain inevitably contain noise due to the domain gap. To tackle this issue, we advance a MetaCorrection framework, where a Domain-aware Meta-learning strategy is devised to benefit Loss Correction (DMLC) for UDA semantic segmentation. In particular, we model the noise distribution of pseudo labels in target domain by introducing a noise transition matrix (NTM) and construct meta data set with domain-invariant source data to guide the estimation of NTM. Through the risk minimization on the meta data set, the optimized NTM thus can correct the noisy issues in pseudo labels and enhance the generalization ability of the model on the target data. Considering the capacity gap between shallow and deep features, we further employ the proposed DMLC strategy to provide matched and compatible supervision signals for different level features, thereby ensuring deep adaptation. Extensive experimental results highlight the effectiveness of our method <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">a</sup> against existing state-of-the-art methods on three benchmarks.
Read moreInformation Maximizing Adaptation Network With Label Distribution Priors for Unsupervised Domain Adaptation
Unsupervised domain adaptation, which transfers knowledge from the source domain to the target domain, has still been a challenging problem. However, previous domain adaptation methods typically minimize the domain discrepancy by using the pseudo target labels. Since the pseudo labels can be noisy, which may cause misalignment and unsatisfying adaptation performance. To address the above challenges, we propose an information maximization adaptation network with label distribution priors. We revisit feature alignment in unsupervised domain adaptation from the perspective of distribution alignment, and find that learning discriminant feature representation requires to minimizing distribution discrepancy and maximizing source mutual information between the outputs of the classifier and feature representations. Due to domain shift, maximizing target mutual information may align features to incorrect class directly. We propose a weighted target mutual information by re-weighting the estimated mutual information via the mean prediction confidence in mini-batch, which can eliminate the negative impact of inaccurate estimation. In addition, we introduce a regularization term of label priors distribution to encourage the similarity to the real label distribution. Extensive experimental results on three benchmark datasets show that our proposed method can achieve remarkable results compared with previous methods.
Read morePseudo Labels for Unsupervised Domain Adaptation: A Review
Conventional machine learning relies on two presumptions: (1) the training and testing datasets follow the same independent distribution, and (2) an adequate quantity of samples is essential for achieving optimal model performance during training. Nevertheless, meeting these two assumptions can be challenging in real-world scenarios. Domain adaptation (DA) is a subfield of transfer learning that focuses on reducing the distribution difference between the source domain (Ds) and target domain (Dt) and subsequently applying the knowledge gained from the Ds task to the Dt task. The majority of current DA methods aim to achieve domain invariance by aligning the marginal probability distributions of the Ds. and Dt. Recent studies have pointed out that aligning marginal probability distributions alone is not sufficient and that alignment of conditional probability distributions is equally important for knowledge migration. Nonetheless, unsupervised DA presents a more significant difficulty in aligning the conditional probability distributions because of the unavailability of labels for the Dt. In response to this issue, there have been several proposed methods by researchers, including pseudo-labeling, which offer novel solutions to tackle the problem. In this paper, we systematically analyze various pseudo-labeling algorithms and their applications in unsupervised DA. First , we summarize the pseudo-label generation methods based on the single and multiple classifiers and actions taken to deal with the problem of imbalanced samples. Second, we investigate the application of pseudo-labeling in category feature alignment and improving feature discrimination. Finally, we point out the challenges and trends of pseudo-labeling algorithms. As far as we know, this article is the initial review of pseudo-labeling techniques for unsupervised DA.
Read moreSF-City: A Source-Free Domain Adaptation Method for City-scale Point Cloud Semantic Segmentation
City-scale point cloud semantic segmentation is an important yet challenging task. Despite progress, existing methods rely heavily on point-wise annotations. An alternative solution is to apply the Unsupervised Domain Adaptation (UDA) approach. Recently, the 2D foundation model has achieved significant progress with training with internet-scale images. Therefore, adapting 2D foundation models to 3D City-scale point clouds is an attempting idea. Due to the data protection and storage issue, 2D source domain data is typically unavailable. Thus, we focus on Source-Free Domain Adaptation (SFDA) and propose a Source-Free City-scale point cloud semantic segmentation method, namely SF-City. Our method leverages knowledge from 2D pre-trained models to generate point-wise pseudo labels for training a 3D semantic segmentation network. We convert point clouds into remote-sensing-like images using Bird's-Eye-View (BEV) projection. However, directly using source models for pseudo label generation is hindered by domain gaps such as viewpoint variations, concept divergences, and geometry loss. To tackle these problems, we propose a Multi-scale Content Feature Extractor (MCFE) to extract holistic and contextual feature representations. Then, an Uncertainty-guided Inter-Model Feature Integrator (UIFI) is introduced to integrate inherent knowledge across source models. Furthermore, the Geometric-guided Pseudo Label Generator (GPLG) is leveraged to introduce geometric information to regulate pseudo labels. Through extensive experiments on two public benchmarks, SF-City demonstrates superior performance, achieving an mIoU of 28.8% on the SensatUrban dataset, outperforming recent state-of-the-art methods CLIPFO3D by about 6.3%.
Read morePseudo-labeling Integrating Centers and Samples with Consistent Selection Mechanism for Unsupervised Domain Adaptation
Pseudo-labeling Integrating Centers and Samples with Consistent Selection Mechanism for Unsupervised Domain Adaptation
Minimizing Outputs’ Differences of Classifiers with Different Responsibilities for Unsupervised Adversarial Domain Adaptation
The adversarial domain adaptation has made many achievements in the field of Unsupervised Domain Adaptation (UDA). However, initial adversarial domain adaptation methods do not consider the category-level domain adaptation, causing confusion between different categories in the different domains. This paper propose a UDA approach called Minimizing Outputs' Differences of Classifiers with Different Responsibilities (MODCDR) to improve this situation. Its framework consists of one generator and two different task-specific classifiers, one working on the source domain, while the other working on the target domain. By training the feature generator to minimize the outputs' discrepancy of these two classifiers, this method not only generates discriminative features, but also achieves the category-level feature distribution alignment simultaneously. In addition, these two classifiers can also produce and filter the reliable pseudo-labels on the target domain, being used for assisting our training. This paper tested this method on three publicly available transfer learning object recognition datasets, with comparison to recent unsupervised adversarial domain adaptation approaches. The results show that the proposed method has superior performance.
Read moreUnsupervised Cross-Scene Aerial Image Segmentation via Spectral Space Transferring and Pseudo-Label Revising
Unsupervised domain adaptation (UDA) is essential since manually labeling pixel-level annotations is consuming and expensive. Since the domain discrepancies have not been well solved, existing UDA approaches yield poor performance compared with supervised learning approaches. In this paper, we propose a novel sequential learning network (SLNet) for unsupervised cross-scene aerial image segmentation. The whole system is decoupled into two sequential parts—the image translation model and segmentation adaptation model. Specifically, we introduce the spectral space transferring (SST) approach to narrow the visual discrepancy. The high-frequency components between the source images and the translated images can be transferred in the Fourier spectral space for better preserving the important identity and fine-grained details. To further alleviate the distribution discrepancy, an efficient pseudo-label revising (PLR) approach was developed to guide pseudo-label learning via entropy minimization. Without additional parameters, the entropy map works as the adaptive threshold, constantly revising the pseudo labels for the target domain. Furthermore, numerous experiments for single-category and multi-category UDA segmentation demonstrate that our SLNet is the state-of-the-art.
Read moreRevisiting Domain-Adaptive Semantic Segmentation via Knowledge Distillation.
Numerous methods for unsupervised domain adaptation (UDA) have been proposed in semantic segmentation, achieving remarkable improvements. These methods are categorized into an adversarial learning-based approach that utilizes an additional discriminator and image translation model, and a self-supervised approach that uses a teacher model to generate pseudo labels. Among them, the self-supervised UDA approaches based on a self-training show excellent adaptability in semantic segmentation. However, erroneous estimates of the pseudo ground truths (PGTs) used in the self-training may often lead to inaccurate updates in the teacher model. Although several attempts have been made to address this issue, the teacher model updated through exponential moving average (EMA) still has a risk of propagating inaccuracies from the PGTs. Inspired by the fact that UDA shares similar principles with knowledge distillation (KD), we revisit the self-training based UDA approach from the perspective of KD and propose a novel UDA approach that employs two different teacher models. Specifically, we utilize both an EMA-updated teacher model to generate PGTs and a frozen teacher model pretrained with source data to transfer knowledge on a feature space. Since the frozen teacher model has no constraint on the model architecture unlike the EMA updated teacher model, we can effectively leverage a better representation power from the larger frozen teacher. Extensive experiments on various backbones (DeepLab-V2 [40] and DAFormer [73]) and scenarios (GTA5 → Cityscapes and SYNTHIA → Cityscapes) show that the proposed method improves segmentation performance in the target domain with its scalability. In particular, our method achieves comparable or better performance than state-of-the-arts even with a lightweight backbone.
Read moreUnsupervised structure subdomain adaptation based the Contrastive Cluster Center for bearing fault diagnosis
Unsupervised structure subdomain adaptation based the Contrastive Cluster Center for bearing fault diagnosis