Robust and Discriminative Self-Taught Learning

Hua Wang, Feiping Nie, Heng Huang
Proceedings of the 30th International Conference on Machine Learning, PMLR 28(3):298-306, 2013.

Abstract

The lack of training data is a common challenge in many machine learning problems, which is often tackled by semi-supervised learning methods or transfer learning methods. The former requires unlabeled images from the same distribution as the labeled ones and the latter leverages labeled images from related homogenous tasks. However, these restrictions often cannot be satisfied. To address this, we propose a novel robust and discriminative self-taught learning approach to utilize any unlabeled data without the above restrictions. Our new approach employs a robust loss function to learn the dictionary, and enforces the structured sparse regularization to automatically select the optimal dictionary basis vectors and incorporate the supervision information contained in the labeled data. We derive an efficient iterative algorithm to solve the optimization problem and rigorously prove its convergence. Promising results in extensive experiments have validated the proposed approach.

Cite this Paper

BibTeX
@InProceedings{pmlr-v28-wang13g, title = {Robust and Discriminative Self-Taught Learning}, author = {Wang, Hua and Nie, Feiping and Huang, Heng}, booktitle = {Proceedings of the 30th International Conference on Machine Learning}, pages = {298--306}, year = {2013}, editor = {Dasgupta, Sanjoy and McAllester, David}, volume = {28}, number = {3}, series = {Proceedings of Machine Learning Research}, address = {Atlanta, Georgia, USA}, month = {17--19 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v28/wang13g.pdf}, url = {https://proceedings.mlr.press/v28/wang13g.html}, abstract = {The lack of training data is a common challenge in many machine learning problems, which is often tackled by semi-supervised learning methods or transfer learning methods. The former requires unlabeled images from the same distribution as the labeled ones and the latter leverages labeled images from related homogenous tasks. However, these restrictions often cannot be satisfied. To address this, we propose a novel robust and discriminative self-taught learning approach to utilize any unlabeled data without the above restrictions. Our new approach employs a robust loss function to learn the dictionary, and enforces the structured sparse regularization to automatically select the optimal dictionary basis vectors and incorporate the supervision information contained in the labeled data. We derive an efficient iterative algorithm to solve the optimization problem and rigorously prove its convergence. Promising results in extensive experiments have validated the proposed approach.} }
Endnote
%0 Conference Paper %T Robust and Discriminative Self-Taught Learning %A Hua Wang %A Feiping Nie %A Heng Huang %B Proceedings of the 30th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2013 %E Sanjoy Dasgupta %E David McAllester %F pmlr-v28-wang13g %I PMLR %P 298--306 %U https://proceedings.mlr.press/v28/wang13g.html %V 28 %N 3 %X The lack of training data is a common challenge in many machine learning problems, which is often tackled by semi-supervised learning methods or transfer learning methods. The former requires unlabeled images from the same distribution as the labeled ones and the latter leverages labeled images from related homogenous tasks. However, these restrictions often cannot be satisfied. To address this, we propose a novel robust and discriminative self-taught learning approach to utilize any unlabeled data without the above restrictions. Our new approach employs a robust loss function to learn the dictionary, and enforces the structured sparse regularization to automatically select the optimal dictionary basis vectors and incorporate the supervision information contained in the labeled data. We derive an efficient iterative algorithm to solve the optimization problem and rigorously prove its convergence. Promising results in extensive experiments have validated the proposed approach.
RIS
TY - CPAPER TI - Robust and Discriminative Self-Taught Learning AU - Hua Wang AU - Feiping Nie AU - Heng Huang BT - Proceedings of the 30th International Conference on Machine Learning DA - 2013/05/26 ED - Sanjoy Dasgupta ED - David McAllester ID - pmlr-v28-wang13g PB - PMLR DP - Proceedings of Machine Learning Research VL - 28 IS - 3 SP - 298 EP - 306 L1 - http://proceedings.mlr.press/v28/wang13g.pdf UR - https://proceedings.mlr.press/v28/wang13g.html AB - The lack of training data is a common challenge in many machine learning problems, which is often tackled by semi-supervised learning methods or transfer learning methods. The former requires unlabeled images from the same distribution as the labeled ones and the latter leverages labeled images from related homogenous tasks. However, these restrictions often cannot be satisfied. To address this, we propose a novel robust and discriminative self-taught learning approach to utilize any unlabeled data without the above restrictions. Our new approach employs a robust loss function to learn the dictionary, and enforces the structured sparse regularization to automatically select the optimal dictionary basis vectors and incorporate the supervision information contained in the labeled data. We derive an efficient iterative algorithm to solve the optimization problem and rigorously prove its convergence. Promising results in extensive experiments have validated the proposed approach. ER -
APA
Wang, H., Nie, F. & Huang, H.. (2013). Robust and Discriminative Self-Taught Learning. Proceedings of the 30th International Conference on Machine Learning, in Proceedings of Machine Learning Research 28(3):298-306 Available from https://proceedings.mlr.press/v28/wang13g.html.

Related Material