2019

Variational Adversarial Active Learning

Sinha, Samarth, Ebrahimi, Sayna, Darrell, Trevor

Understand

Active learning aims to develop label-efficient algorithms by sampling the most representative queries to be labeled by an oracle.

We describe a pool-based semi-supervised active learning algorithm that implicitly learns this sampling mechanism in an adversarial manner.
Unlike conventional active learning algorithms, our approach is task agnostic, i.e., it does not depend on the performance of the task for which we are trying to acquire labeled data.
Our method learns a latent space using a variational autoencoder (VAE) and an adversarial network trained to discriminate between unlabeled and labeled data.

Built on

A mathematical theory of communication
Claude Elwood Shannon · 1948
Earlier work this paper cites.
Elementary applied statistics: for students in behavioral science
Linton C Freeman · 1965
Earlier work this paper cites.
A class of wasserstein metrics for probability distributions
Clark R Givens, Rae Michael Shortt, et al · 1984
Earlier work this paper cites.
Information-based objective functions for active data selection
David JC MacKay · 1992
Earlier work this paper cites.
An introduction to the bootstrap
Bradley Efron and Robert J Tibshirani · 1994
Earlier work this paper cites.
Active learning with statistical models
David A Cohn, Zoubin Ghahramani, and Michael I Jordan · 1996
Earlier work this paper cites.
Selective sampling using the query by committee algorithm
Yoav Freund, H Sebastian Seung, Eli Shamir, and Naftali Tishby · 1997
Earlier work this paper cites.
Employing em and pool-based active learning for text classification
Andrew Kachites McCallumzy and Kamal Nigamy · 1998
Earlier work this paper cites.
Toward optimal active learning through monte carlo estimation of error reduction
Nicholas Roy and Andrew McCallum · 2001
Earlier work this paper cites.
Support vector machine active learning with applications to text classification
Simon Tong and Daphne Koller · 2001
Earlier work this paper cites.
Incorporating diversity in active learning with support vector machines
Klaus Brinker · 2003
Earlier work this paper cites.
Diverse ensembles for active learning
Prem Melville and Raymond J Mooney · 2004
Earlier work this paper cites.
Active learning using pre-clustering
Hieu T Nguyen and Arnold Smeulders · 2004
Earlier work this paper cites.
Query by committee made real
Ran Gilad-Bachrach, Amir Navot, and Naftali Tishby · 2006
Earlier work this paper cites.
Caltech-256 object category dataset
Gregory Griffin, Alex Holub, and Pietro Perona · 2007
Earlier work this paper cites.
Active learning with gaussian processes for object categorization
Ashish Kapoor, Kristen Grauman, Raquel Urtasun, and Trevor Darrell · 2007
Earlier work this paper cites.
High-dimensional data analysis
Damien François · 2008
Earlier work this paper cites.
ImageNet: A Large-Scale Hierarchical Image Database
Jia Deng, Wei Dong, Richard Socher, Lie-Jia Li, Kai Li, and Li Fei-Fei · 2009
Earlier work this paper cites.
Learning multiple layers of features from tiny images
Alex Krizhevsky and Geoffrey Hinton · 2009
Earlier work this paper cites.
Understanding the difficulty of training deep feedforward neural networks
Xavier Glorot and Yoshua Bengio · 2010
Earlier work this paper cites.

Similar

Active learning
Burr Settles · 2012
Cited alongside, same era.
Representation learning: A review and new perspectives
Yoshua Bengio, Aaron Courville, and Pascal Vincent · 2013
Cited alongside, same era.
Auto-encoding variational bayes
Original
Diederik P Kingma and Max Welling · 2013
Cited alongside, same era.
Adaptive active learning for image classification
Xin Li and Yuhong Guo · 2013
Cited alongside, same era.
Generative adversarial nets
Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio · 2014
Cited alongside, same era.
Active learning literature survey. 2010
Burr Settles · 2014
Cited alongside, same era.
Very deep convolutional networks for large-scale image recognition
Original
Karen Simonyan and Andrew Zisserman · 2014
Cited alongside, same era.
Adam: A method for stochastic optimization
Diederik P Kingma and Jimmy Ba · 2015
Cited alongside, same era.
Adversarial autoencoders
Original
Alireza Makhzani, Jonathon Shlens, Navdeep Jaitly, Ian Goodfellow, and Brendan Frey · 2015
Cited alongside, same era.
Learning structured output representation using deep conditional generative models
Kihyuk Sohn, Honglak Lee, and Xinchen Yan · 2015
Cited alongside, same era.
Querying discriminative and representative samples for batch mode active learning
Zheng Wang and Jieping Ye · 2015
Cited alongside, same era.
The cityscapes dataset for semantic urban scene understanding
Marius Cordts, Mohamed Omran, Sebastian Ramos, Timo Rehfeld, Markus Enzweiler, Rodrigo Benenson, Uwe Franke, Stefan Roth, and Bernt Schiele · 2016
Cited alongside, same era.

Then

Towards deep learning models resistant to adversarial attacks
Original
Aleksander Madry, Aleksandar Makelov, Ludwig Schmidt, Dimitris Tsipras, and Adrian Vladu · 2017
Later among the works it cites.
Adversarial variational bayes: Unifying variational autoencoders and generative adversarial networks
Lars Mescheder, Sebastian Nowozin, and Andreas Geiger · 2017
Later among the works it cites.
Wasserstein auto-encoders
Original
Ilya Tolstikhin, Olivier Bousquet, Sylvain Gelly, and Bernhard Schoelkopf · 2017
Later among the works it cites.
Ensemble adversarial training: Attacks and defenses
Original
Florian Tramèr, Alexey Kurakin, Nicolas Papernot, Ian Goodfellow, Dan Boneh, and Patrick McDaniel · 2017
Later among the works it cites.
Adversarial discriminative domain adaptation
Eric Tzeng, Judy Hoffman, Kate Saenko, and Trevor Darrell · 2017
Later among the works it cites.
Cost-effective active learning for deep image classification
Keze Wang, Dongyu Zhang, Ya Li, Ruimao Zhang, and Liang Lin · 2017
Later among the works it cites.
Suggestive annotation: A deep active learning framework for biomedical image segmentation
Lin Yang, Yizhe Zhang, Jianxu Chen, Siyuan Zhang, and Danny Z Chen · 2017
Later among the works it cites.
Dilated residual networks
Fisher Yu, Vladlen Koltun, and Thomas A Funkhouser · 2017
Later among the works it cites.
Generative adversarial active learning
Original
Jia-Jie Zhu and José Bento · 2017
Later among the works it cites.
The power of ensembles for active learning in image classification
William H Beluch, Tim Genewein, Andreas Nürnberger, and Jan M Köhler · 2018
Later among the works it cites.
Adversarial active learning for sequences labeling and generation
Yue Deng, KaWai Chen, Yilin Shen, and Hongxia Jin · 2018
Later among the works it cites.
Disentangling by factorising
Original
Hyunjik Kim and Andriy Mnih · 2018
Later among the works it cites.
Cost-sensitive active learning for intracranial hemorrhage detection
Weicheng Kuo, Christian Häne, Esther Yuh, Pratik Mukherjee, and Jitendra Malik · 2018
Later among the works it cites.
Efficient active learning for image classification and segmentation using a sample selection and conditional generative adversarial network
Dwarikanath Mahapatra, Behzad Bozorgtabar, Jean-Philippe Thiran, and Mauricio Reyes · 2018
Later among the works it cites.
Adversarial sampling for active learning
Original
Christoph Mayer and Radu Timofte · 2018
Later among the works it cites.
Active learning for convolutional neural networks: A core-set approach
Ozan Sener and Silvio Savarese · 2018
Later among the works it cites.
Bdd100k: A diverse driving video database with scalable annotation tooling
Original
Fisher Yu, Wenqi Xian, Yingying Chen, Fangchen Liu, Mike Liao, Vashisht Madhavan, and Trevor Darrell · 2018
Later among the works it cites.
Uncertainty-guided continual learning with bayesian neural networks
Original
Sayna Ebrahimi, Mohamed Elhoseiny, Trevor Darrell, and Marcus Rohrbach · 2019
Closest in time.
Generalized zero-and few-shot learning via aligned variational autoencoders
Edgar Schonfeld, Sayna Ebrahimi, Samarth Sinha, Trevor Darrell, and Zeynep Akata · 2019
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…