Unsupervised Learning by Predicting Noise

Piotr Bojanowski, Armand Joulin
Proceedings of the 34th International Conference on Machine Learning, PMLR 70:517-526, 2017.

Abstract

Convolutional neural networks provide visual features that perform remarkably well in many computer vision applications. However, training these networks requires significant amounts of supervision; this paper introduces a generic framework to train such networks, end-to-end, with no supervision. We propose to fix a set of target representations, called Noise As Targets (NAT), and to constrain the deep features to align to them. This domain agnostic approach avoids the standard unsupervised learning issues of trivial solutions and collapsing of the features. Thanks to a stochastic batch reassignment strategy and a separable square loss function, it scales to millions of images. The proposed approach produces representations that perform on par with the state-of-the-arts among unsupervised methods on ImageNet and Pascal VOC.

Cite this Paper


BibTeX
@InProceedings{pmlr-v70-bojanowski17a, title = {Unsupervised Learning by Predicting Noise}, author = {Piotr Bojanowski and Armand Joulin}, booktitle = {Proceedings of the 34th International Conference on Machine Learning}, pages = {517--526}, year = {2017}, editor = {Precup, Doina and Teh, Yee Whye}, volume = {70}, series = {Proceedings of Machine Learning Research}, month = {06--11 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v70/bojanowski17a/bojanowski17a.pdf}, url = { http://proceedings.mlr.press/v70/bojanowski17a.html }, abstract = {Convolutional neural networks provide visual features that perform remarkably well in many computer vision applications. However, training these networks requires significant amounts of supervision; this paper introduces a generic framework to train such networks, end-to-end, with no supervision. We propose to fix a set of target representations, called Noise As Targets (NAT), and to constrain the deep features to align to them. This domain agnostic approach avoids the standard unsupervised learning issues of trivial solutions and collapsing of the features. Thanks to a stochastic batch reassignment strategy and a separable square loss function, it scales to millions of images. The proposed approach produces representations that perform on par with the state-of-the-arts among unsupervised methods on ImageNet and Pascal VOC.} }
Endnote
%0 Conference Paper %T Unsupervised Learning by Predicting Noise %A Piotr Bojanowski %A Armand Joulin %B Proceedings of the 34th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2017 %E Doina Precup %E Yee Whye Teh %F pmlr-v70-bojanowski17a %I PMLR %P 517--526 %U http://proceedings.mlr.press/v70/bojanowski17a.html %V 70 %X Convolutional neural networks provide visual features that perform remarkably well in many computer vision applications. However, training these networks requires significant amounts of supervision; this paper introduces a generic framework to train such networks, end-to-end, with no supervision. We propose to fix a set of target representations, called Noise As Targets (NAT), and to constrain the deep features to align to them. This domain agnostic approach avoids the standard unsupervised learning issues of trivial solutions and collapsing of the features. Thanks to a stochastic batch reassignment strategy and a separable square loss function, it scales to millions of images. The proposed approach produces representations that perform on par with the state-of-the-arts among unsupervised methods on ImageNet and Pascal VOC.
APA
Bojanowski, P. & Joulin, A.. (2017). Unsupervised Learning by Predicting Noise. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:517-526 Available from http://proceedings.mlr.press/v70/bojanowski17a.html .

Related Material