Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
By Kuniaki Saito, Kohei Watanabe, Yoshitaka Ushiku, Tatsuya Harada. 2017
2,008 citations recorded by the source corpus.
In this work, we present a method for unsupervised domain adaptation. Many adversarial learning methods train domain classifier networks to distinguish the features as either a source or target and train a feature generator network to mimic the discriminator. Two problems exist with these methods. First, the domain classifier only tries to distinguish the features as a source or target and thus does not consider task-specific decision boundaries between classes. Therefore, a trained generator can generate ambiguous features near class boundaries. Second, these methods aim to completely match the feature distributions between different domains, which is difficult because of each domain's characteristics. To solve these problems, we introduce a new approach that attempts to align distributions of source and target by utilizing the task-specific decision boundaries. We propose to maximize the discrepancy between two classifiers' outputs to detect target samples that are far from the support of the source. A feature generator learns to generate target features near the support to minimize the discrepancy. Our method outperforms other methods on several datasets of image classification and semantic segmentation. The codes are available at https://github.com/mil-tokyo/MCD_DA
Subfields: Artificial Intelligence, Health Informatics, Radiology, Nuclear Medicine and Imaging
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Health Informatics
In Radiology, Nuclear Medicine and Imaging
Citations in the corpus
- Prototype-Guided Continual Adaptation for Class-Incremental Unsupervised Domain Adaptation (2022, methodology): “C→I Step 1 [0, 2, 7, 9, 11, 27, 28, 29, 30, 33] Step 2 [37, 39, 40, 44, 45, 47, 50, 60, 62, 68] Step 3 [71, 75, 76, 82, 85, 86, 87, 88, 89, 90] Step 4 [92, 94, 96, 97, 106, 107, 108, 109, 110, 112] Step 5 [114, 115, 116, 123, 126, 128, 133, 134, 141, 145] Step 6 [146, 150, 151, 157, 160, 163, 165, 1”
- Unsupervised Model Adaptation for Continual Semantic Segmentation (2020, methodology): “A large group of the existing UDA algorithms for image segmentation use adversarial learning for extracting domaininvariant features (Luc et al. 2016; Bousmalis et al. 2017; Hoffman et al. 2018b; Murez et al. 2018; Saito et al. 2018; Sankaranarayanan et al. 2018; Dhouib, Redko, and Lartizien 2020).”
- Moment Matching for Multi-Source Domain Adaptation (2018, methodology): “Single-Source Adaptation To demonstrate the intrinsic difficulty of DomainNet, we evaluate multiple state-ofthe-art algorithms for single-source adaptation: Deep Alignment Network (DAN) [25], Joint Adaptation Network (JAN) [27], Domain Adversarial Neural Network (DANN) [8], Residual Transfer Network”
- Class-Incremental Domain Adaptation (2020, neutral): “However, most UDA methods [10,12,44,52] assume that the two domains share the same label space (as shown in Fig.”
- Contrastive Adaptation Network for Unsupervised Domain Adaptation (2019, methodology): “7 MCD [32] 87.”
- ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation (2018, methodology): “[32, 33] propose another interesting way of using adversarial training: They get two predictions on the target domain image, this is done either by two classifiers [33] or using dropout in the classifier [32].”
- Unsupervised Domain Adaptation for Speech Recognition via Uncertainty Driven Self-Training (2020, background): “versal layer (GRL) [10], and training using discrepancy losses [11].”
- Learning Semantic Representations for Unsupervised Domain Adaptation (2018, methodology): “Recently, adversarial learning has been widely adopted in domain adaptation (Ganin & Lempitsky, 2015; Tzeng et al., 2015; Hoffman et al., 2017; Motiian et al., 2017a; Tzeng et al., 2017; Saito et al., 2017b; Long et al., 2017a; Luo et al., 2017; Sankaranarayanan et al.).”
- Towards Discriminability and Diversity: Batch Nuclear-Norm Maximization Under Label Insufficient Situations (2020, background): “Domain adaptation [18, 39, 26, 56] reduces the domain discrepancy between labeled and unlabeled examples.”
- Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation (2021, methodology): “Finally, BAIT [56] extends MCD [35] to the SFDA”
- Unsupervised Intra-Domain Adaptation for Semantic Segmentation Through Self-Supervision (2020, background): “Recently, adversarial-based UDA approaches have shown great capabilities in learning domain invariant features, even for complex tasks like semantic segmentation [28, 4, 26, 25, 21, 11, 17].”
- Domain-Specific Batch Normalization for Unsupervised Domain Adaptation (2019, methodology): “Implementation details Following [3, 20], as the backbone networks of our framework, we adopt ResNet-101 for the VisDA-C dataset and ResNet-50 for Office-31 and Office-Home datasets.”
- Category Anchor-Guided Unsupervised Domain Adaptation for Semantic Segmentation (2019, background): “tried to align the feature distributions and learn discriminative domain-invariant features by utilizing task-specific classifiers as a discriminator [33].”
- Model Adaptation: Unsupervised Domain Adaptation Without Source Data (2020, background): “In [49], instead of relying on a domain discriminator, Saito et al.”
- Deep Visual Domain Adaptation: A Survey (2018, background): “In [97] , two classifiers are treated s discriminators and are trained to maximize the discrepancy to etect target samples outside the support of the source, while a eature extractor is trained to minimize the discrepancy by generting target features near the support.”
- Deep Unsupervised Domain Adaptation: A Review of Recent Advances and Perspectives (2022, background): “In Saito et al. (2018), there are two discriminators to maximize the discrepancy of each class in the target domain, which renders the target domain features to have a wider class-wise boundary region to facilitate the classification.”
- Domain2Vec: Domain Embedding for Unsupervised Domain Adaptation (2020, methodology): “For both single best and source combine experiment setting, we take the following methods as our baselines: Deep Alignment Network (DAN) [4], Joint Adaptation Network (JAN) [2], Domain Adversarial Neural Network (DANN) [12], Residual Transfer Network (RTN) [69], Adversarial Deep Domain Adaptation (A”
- A Survey of Unsupervised Deep Domain Adaptation (2018, methodology): “2 99.6 VADA[220]b∗ 94.5 73.3 95.7 94.9 99.2 DeepJDOT[50] 95.7 96.4 96.7 92.4 CyCADA[96]∗ 95.6 ±0.2 96.5 ±0.1 90.4 ±0.4 Gen. to Adapt[208]∗ 92.8 ±0.9 90.8 ±1.3 92.4 ±0.9 SimNet[187]∗ 96.4 95.6 90.5 MCD[206]∗ 96.5 ±0.3 94.1 ±0.3 96.2 ±0.4 94.4 ±0.3 GAGL[250]b∗ 96.7 74.6 94.9 93.1 97.6 SBADA-GAN[201]b∗”
- Denoised Maximum Classifier Discrepancy for Source-Free Unsupervised Domain Adaptation (2022, methodology): “While many methods have been proposed to solve the UDA problem (Ganin and Lempitsky 2015; Long et al. 2018; Saito et al. 2018; Dong et al. 2020; Liu et al. 2021; Deng et al. 2021; Dong et al. 2021), these approaches is needed to access the source domain data during the training process.”