Learning To Stop While Learning To Predict

Xinshi Chen, Hanjun Dai, Yu Li, Xin Gao, Le Song
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:1520-1530, 2020.

Abstract

There is a recent surge of interest in designing deep architectures based on the update steps in traditional algorithms, or learning neural networks to improve and replace traditional algorithms. While traditional algorithms have certain stopping criteria for outputting results at different iterations, many algorithm-inspired deep models are restricted to a “fixed-depth” for all inputs. Similar to algorithms, the optimal depth of a deep architecture may be different for different input instances, either to avoid “over-thinking”, or because we want to compute less for operations converged already. In this paper, we tackle this varying depth problem using a steerable architecture, where a feed-forward deep model and a variational stopping policy are learned together to sequentially determine the optimal number of layers for each input instance. Training such architecture is very challenging. We provide a variational Bayes perspective and design a novel and effective training procedure which decomposes the task into an oracle model learning stage and an imitation stage. Experimentally, we show that the learned deep model along with the stopping policy improves the performances on a diverse set of tasks, including learning sparse recovery, few-shot meta learning, and computer vision tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-chen20c, title = {Learning To Stop While Learning To Predict}, author = {Chen, Xinshi and Dai, Hanjun and Li, Yu and Gao, Xin and Song, Le}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {1520--1530}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/chen20c/chen20c.pdf}, url = {https://proceedings.mlr.press/v119/chen20c.html}, abstract = {There is a recent surge of interest in designing deep architectures based on the update steps in traditional algorithms, or learning neural networks to improve and replace traditional algorithms. While traditional algorithms have certain stopping criteria for outputting results at different iterations, many algorithm-inspired deep models are restricted to a “fixed-depth” for all inputs. Similar to algorithms, the optimal depth of a deep architecture may be different for different input instances, either to avoid “over-thinking”, or because we want to compute less for operations converged already. In this paper, we tackle this varying depth problem using a steerable architecture, where a feed-forward deep model and a variational stopping policy are learned together to sequentially determine the optimal number of layers for each input instance. Training such architecture is very challenging. We provide a variational Bayes perspective and design a novel and effective training procedure which decomposes the task into an oracle model learning stage and an imitation stage. Experimentally, we show that the learned deep model along with the stopping policy improves the performances on a diverse set of tasks, including learning sparse recovery, few-shot meta learning, and computer vision tasks.} }
Endnote
%0 Conference Paper %T Learning To Stop While Learning To Predict %A Xinshi Chen %A Hanjun Dai %A Yu Li %A Xin Gao %A Le Song %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-chen20c %I PMLR %P 1520--1530 %U https://proceedings.mlr.press/v119/chen20c.html %V 119 %X There is a recent surge of interest in designing deep architectures based on the update steps in traditional algorithms, or learning neural networks to improve and replace traditional algorithms. While traditional algorithms have certain stopping criteria for outputting results at different iterations, many algorithm-inspired deep models are restricted to a “fixed-depth” for all inputs. Similar to algorithms, the optimal depth of a deep architecture may be different for different input instances, either to avoid “over-thinking”, or because we want to compute less for operations converged already. In this paper, we tackle this varying depth problem using a steerable architecture, where a feed-forward deep model and a variational stopping policy are learned together to sequentially determine the optimal number of layers for each input instance. Training such architecture is very challenging. We provide a variational Bayes perspective and design a novel and effective training procedure which decomposes the task into an oracle model learning stage and an imitation stage. Experimentally, we show that the learned deep model along with the stopping policy improves the performances on a diverse set of tasks, including learning sparse recovery, few-shot meta learning, and computer vision tasks.
APA
Chen, X., Dai, H., Li, Y., Gao, X. & Song, L.. (2020). Learning To Stop While Learning To Predict. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:1520-1530 Available from https://proceedings.mlr.press/v119/chen20c.html.

Related Material