Stochastic Adversarial Gradient Embedding for Active Domain Adaptation

Unsupervised Domain Adaptation (UDA) aims to bridge the gap between a source domain, where labelled data are available, and a target domain only represented with unlabelled data. If domain invariant representations have dramatically improved the adaptability of models, to guarantee their good transferability remains a challenging problem. This paper addresses this problem by using active learning to annotate a small budget of target data. Although this setup, called Active Domain Adaptation (ADA), deviates from UDA's standard setup, a wide range of practical applications are faced with this situation. To this purpose, we introduce \textit{Stochastic Adversarial Gradient Embedding} (SAGE), a framework that makes a triple contribution to ADA. First, we select for annotation target samples that are likely to improve the representations' transferability by measuring the variation, before and after annotation, of the transferability loss gradient. Second, we increase sampling diversity by promoting different gradient directions. Third, we introduce a novel training procedure for actively incorporating target samples when learning invariant representations. SAGE is based on solid theoretical ground and validated on various UDA benchmarks against several baselines. Our empirical investigation demonstrates that SAGE takes the best of uncertainty \textit{vs} diversity samplings and improves representations transferability substantially.

[1]  Junzhou Huang,et al.  Progressive Feature Alignment for Unsupervised Domain Adaptation , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Koby Crammer,et al.  A theory of learning from different domains , 2010, Machine Learning.

[3]  Dan Wang,et al.  A new active labeling method for deep learning , 2014, 2014 International Joint Conference on Neural Networks (IJCNN).

[4]  MarchandMario,et al.  Domain-adversarial training of neural networks , 2016 .

[5]  Hsuan-Tien Lin,et al.  Active Learning by Learning , 2015, AAAI.

[6]  Jianmin Wang,et al.  Transferability vs. Discriminability: Batch Spectral Penalization for Adversarial Domain Adaptation , 2019, ICML.

[7]  François Laviolette,et al.  Domain-Adversarial Training of Neural Networks , 2015, J. Mach. Learn. Res..

[8]  Neil D. Lawrence,et al.  Dataset Shift in Machine Learning , 2009 .

[9]  Michael I. Jordan,et al.  Universal Domain Adaptation , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[10]  Sergei Vassilvitskii,et al.  k-means++: the advantages of careful seeding , 2007, SODA '07.

[11]  Daumé,et al.  Domain Adaptation meets Active Learning , 2010, HLT-NAACL 2010.

[12]  John Langford,et al.  Deep Batch Active Learning by Diverse, Uncertain Gradient Lower Bounds , 2019, ICLR.

[13]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[14]  Ivan Laptev,et al.  Learning and Transferring Mid-level Image Representations Using Convolutional Neural Networks , 2014, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[15]  Natalia Gimelshein,et al.  PyTorch: An Imperative Style, High-Performance Deep Learning Library , 2019, NeurIPS.

[16]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[17]  Avishek Saha,et al.  Active Supervised Domain Adaptation , 2011, ECML/PKDD.

[18]  Philippe Very,et al.  Robust Domain Adaptation: Representations, Weights and Inductive Bias , 2020, ECML/PKDD.

[19]  Subhransu Maji,et al.  Active Adversarial Domain Adaptation , 2019, 2020 IEEE Winter Conference on Applications of Computer Vision (WACV).

[20]  Rajesh Ranganath,et al.  Support and Invertibility in Domain-Invariant Representations , 2019, AISTATS.

[21]  Gilles Louppe,et al.  Independent consultant , 2013 .

[22]  Han Zhao,et al.  On Learning Invariant Representations for Domain Adaptation , 2019, ICML.

[23]  Silvio Savarese,et al.  Active Learning for Convolutional Neural Networks: A Core-Set Approach , 2017, ICLR.

[24]  Burr Settles,et al.  Active Learning Literature Survey , 2009 .

[25]  Trevor Darrell,et al.  Adapting Visual Category Models to New Domains , 2010, ECCV.

[26]  Geoffrey J. Gordon,et al.  Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift , 2020, NeurIPS.

[27]  Victor S. Lempitsky,et al.  Unsupervised Domain Adaptation by Backpropagation , 2014, ICML.

[28]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[29]  Yoav Goldberg,et al.  Are We Modeling the Task or the Annotator? An Investigation of Annotator Bias in Natural Language Understanding Datasets , 2019, EMNLP.

[30]  Michael I. Jordan,et al.  Learning Transferable Features with Deep Adaptation Networks , 2015, ICML.

[31]  Jing Zhang,et al.  Importance Weighted Adversarial Nets for Partial Domain Adaptation , 2018, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[32]  Koby Crammer,et al.  Analysis of Representations for Domain Adaptation , 2006, NIPS.

[33]  Yoshua Bengio,et al.  How transferable are features in deep neural networks? , 2014, NIPS.

[34]  Steve Hanneke,et al.  Theory of Disagreement-Based Active Learning , 2014, Found. Trends Mach. Learn..

[35]  David Lopez-Paz,et al.  Invariant Risk Minimization , 2019, ArXiv.

[36]  Gaël Varoquaux,et al.  The NumPy Array: A Structure for Efficient Numerical Computation , 2011, Computing in Science & Engineering.

[37]  Michael I. Jordan,et al.  Deep Transfer Learning with Joint Adaptation Networks , 2016, ICML.

[38]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[39]  Dan Roth,et al.  Margin-Based Active Learning for Structured Output Spaces , 2006, ECML.

[40]  John Langford,et al.  Agnostic active learning , 2006, J. Comput. Syst. Sci..

[41]  Sethuraman Panchanathan,et al.  Joint Transfer and Batch-mode Active Learning , 2013, ICML.

[42]  Michael I. Jordan,et al.  Transferable Adversarial Training: A General Approach to Adapting Deep Classifiers , 2019, ICML.

[43]  Kate Saenko,et al.  VisDA: The Visual Domain Adaptation Challenge , 2017, ArXiv.

[44]  John Schulman,et al.  Concrete Problems in AI Safety , 2016, ArXiv.

[45]  Victor Bouvier,et al.  Target Consistency for Domain Adaptation: when Robustness meets Transferability , 2020, ArXiv.

[46]  Pietro Perona,et al.  Recognition in Terra Incognita , 2018, ECCV.

[47]  Michael I. Jordan,et al.  Unsupervised Domain Adaptation with Residual Transfer Networks , 2016, NIPS.

[48]  Philippe Very,et al.  Hidden Covariate Shift: A Minimal Assumption For Domain Adaptation , 2019, ArXiv.

[49]  Qiang Yang,et al.  A Survey on Transfer Learning , 2010, IEEE Transactions on Knowledge and Data Engineering.

[50]  John D. Hunter,et al.  Matplotlib: A 2D Graphics Environment , 2007, Computing in Science & Engineering.

[51]  Gary Marcus,et al.  The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence , 2020, ArXiv.

[52]  Michael I. Jordan,et al.  Conditional Adversarial Domain Adaptation , 2017, NeurIPS.

[53]  Jianmin Wang,et al.  Partial Adversarial Domain Adaptation , 2018, ECCV.