Domain-Adversarial Training of Neural Networks
By Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pascal Germain, Hugo Larochelle, François Laviolette, Mario Marchand, Victor S. Lempitsky. 2015
11,758 citations recorded by the source corpus.
We introduce a new representation learning approach for domain adaptation, in which data at training and test time come from similar but different distributions. Our approach is directly inspired by the theory on domain adaptation suggesting that, for effective domain transfer to be achieved, predictions must be made based on features that cannot discriminate between the training (source) and test (target) domains. The approach implements this idea in the context of neural network architectures that are trained on labeled data from the source domain and unlabeled data from the target domain (no labeled target-domain data is necessary). As the training progresses, the approach promotes the emergence of features that are (i) discriminative for the main learning task on the source domain and (ii) indiscriminate with respect to the shift between the domains. We show that this adaptation behaviour can be achieved in almost any feed-forward model by augmenting it with few standard layers and a new gradient reversal layer. The resulting augmented architecture can be trained using standard backpropagation and stochastic gradient descent, and can thus be implemented with little effort using any of the deep learning packages. We demonstrate the success of our approach for two distinct classification problems (document sentiment analysis and image classification), where state-of-the-art domain adaptation performance on standard benchmarks is achieved. We also validate the approach for descriptor learning task in the context of person re-identification application.
Subfields: Artificial Intelligence, Biomedical Engineering, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning
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
- Yoshua Bengio
- Mohit Bansal
- Sergey Levine
- Percy Liang
- Quoc V. Le
- Jianfeng Gao
- Dacheng Tao
- Trevor Darrell
In Biomedical Engineering
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Trevor Darrell
- Xiaogang Wang
- Yi Yang
- Jitendra Malik
- Andrew Zisserman
In Representation & Self-Supervised Learning
- Yoshua Bengio
- Trevor Darrell
- Dacheng Tao
- Mingsheng Long
- Mohammad Rostami
- Tao A. Xiang
- Jianmin Wang
- Sergey Levine
Citations in the corpus
- Learning Robust Global Representations by Penalizing Local Predictive Power (2019, methodology): “Via the reverse gradient technique popularized by Ganin et al. (2016), our network is optimized to fool the side classifiers and simultaneously optimized to output correct predictions at the final layer.”
- Unsupervised Image-to-Image Translation Networks (2017, background): “fy samples in a new domain (target domain) where labeled samples in the new domain are unavailable during training. Early works have explored ideas from subspace learning [4] to deep feature learning [5, 17, 26]. We performed multi-task learning where we trained the framework to 1) translate images ”
- Coupled Generative Adversarial Networks (2016, background): “Early works have explored ideas from subspace learning [17, 18] to deep discriminative network learning [19, 20, 21].”
- Toward Domain-Invariant Speech Recognition via Large Scale Training (2018, methodology): “We also expect modeling strategies like domain adversarial training [18] and better training techniques [37] to improve performance.”
- Unsupervised Speech Representation Learning Using WaveNet Autoencoders (2019, background): “enforce invariance to a known nuisance factor [65].”
- Amnesic Probing: Behavioral Explanation with Amnesic Counterfactuals (2020, methodology): “One commonly used method for information removal relies on adversarial training through the gradient reversal layer technique (Ganin et al., 2016).”
- Unsupervised Machine Translation Using Monolingual Corpora Only (2017, methodology): “In order to add such a constraint, we train a neural network, which we will refer to as the discriminator , to classify between the encoding of source sentences and the encoding of target sentences (Ganin et al., 2016).”
- NIPS 2016 Tutorial: Generative Adversarial Networks (2016, neutral): “• Domain adaptation via domain-adversarial learning (Ganin et al., 2015).”
- Unsupervised Cross-Domain Image Generation (2016, background): “Unsupervised domain adaptation addresses the following problem: given a labeled training set in S × Y , for some target space Y , and an unlabeled set of samples from domain T , learn a function h : T → Y (Chen et al., 2012; Ganin et al., 2016).”
- Word Translation Without Parallel Data (2017, positive): “This approach is in line with the work of Ganin et al. (2016), who proposed to learn latent representations invariant to the input domain, where in our case, a domain is represented by a language (source or target).”
- Professor Forcing: A New Algorithm for Training Recurrent Networks (2016, methodology): “We evaluate Professor Forcing on character-level language modeling on Penn-Treebank corpus, which has an alphabet size of 50 and consists of 5059k characters for training, 396k characters for validation and 446k characters for test.”
- Adversarial Discriminative Domain Adaptation (2017, methodology): “Following previous work [19], we focus our evaluation on three domain shifts: amazon to we-bcam ( A → W ), dslr to webcam ( D → W ), and webcam to dslr ( W → D ).”
- Multiple-Source Adaptation with Domain Classifiers (2020, neutral): “…al., 2008, 2009a, Hoffman et al., 2018, Pan and Yang, 2010, Muandet et al., 2013, Xu et al., 2014, Hoffman et al., 2012, Gong et al., 2013a,b, Zhang et al., 2015, Ganin et al., 2016, Tzeng et al., 2015, Motiian et al., 2017b,a, Wang et al., 2019, Konstantinov and Lampert, 2019, Liu et al., 2015].”
- A Theory of Multiple-Source Adaptation with Limited Target Labeled Data (2020, background): “ith only unlabeled target data available and provided generalization bounds for classification and regression. Recently, there have been many experimental studies of domain adaptation in various tasks.Ganin et al. (2016) proposed to learn features that cannot discriminate between source and target do”
- Three Approaches for Personalization with Applications to Federated Learning (2020, background): “Personalization is different from most domain adaptation works as they assume they only have access to unlabeled target data (Mansour et al., 2009a; Ganin et al., 2016; Zhao et al., 2018a), whereas in personalization we have access to labeled target data.”