On the Complexity of Bandit and Derivative-Free Stochastic Convex Optimization

Ohad Shamir
Proceedings of the 26th Annual Conference on Learning Theory, PMLR 30:3-24, 2013.

Abstract

The problem of stochastic convex optimization with bandit feedback (in the learning community) or without knowledge of gradients (in the optimization community) has received much attention in recent years, in the form of algorithms and performance upper bounds. However, much less is known about the inherent complexity of these problems, and there are few lower bounds in the literature, especially for nonlinear functions. In this paper, we investigate the attainable error/regret in the bandit and derivative-free settings, as a function of the dimension d and the available number of queries T. We provide a precise characterization of the attainable performance for strongly-convex and smooth functions, which also imply a non-trivial lower bound for more general problems. Moreover, we prove that in both the bandit and derivative-free setting, the required number of queries must scale at least quadratically with the dimension. Finally, we show that on the natural class of quadratic functions, it is possible to obtain a “fast” O(1/T) error rate in terms of T, under mild assumptions, even without having access to gradients. To the best of our knowledge, this is the first such rate in a derivative-free stochastic setting, and holds despite previous results which seem to imply the contrary.

Cite this Paper

BibTeX
@InProceedings{pmlr-v30-Shamir13, title = {On the Complexity of Bandit and Derivative-Free Stochastic Convex Optimization}, author = {Shamir, Ohad}, booktitle = {Proceedings of the 26th Annual Conference on Learning Theory}, pages = {3--24}, year = {2013}, editor = {Shalev-Shwartz, Shai and Steinwart, Ingo}, volume = {30}, series = {Proceedings of Machine Learning Research}, address = {Princeton, NJ, USA}, month = {12--14 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v30/Shamir13.pdf}, url = {https://proceedings.mlr.press/v30/Shamir13.html}, abstract = {The problem of stochastic convex optimization with bandit feedback (in the learning community) or without knowledge of gradients (in the optimization community) has received much attention in recent years, in the form of algorithms and performance upper bounds. However, much less is known about the inherent complexity of these problems, and there are few lower bounds in the literature, especially for nonlinear functions. In this paper, we investigate the attainable error/regret in the bandit and derivative-free settings, as a function of the dimension d and the available number of queries T. We provide a precise characterization of the attainable performance for strongly-convex and smooth functions, which also imply a non-trivial lower bound for more general problems. Moreover, we prove that in both the bandit and derivative-free setting, the required number of queries must scale at least quadratically with the dimension. Finally, we show that on the natural class of quadratic functions, it is possible to obtain a “fast” O(1/T) error rate in terms of T, under mild assumptions, even without having access to gradients. To the best of our knowledge, this is the first such rate in a derivative-free stochastic setting, and holds despite previous results which seem to imply the contrary.} }
Endnote
%0 Conference Paper %T On the Complexity of Bandit and Derivative-Free Stochastic Convex Optimization %A Ohad Shamir %B Proceedings of the 26th Annual Conference on Learning Theory %C Proceedings of Machine Learning Research %D 2013 %E Shai Shalev-Shwartz %E Ingo Steinwart %F pmlr-v30-Shamir13 %I PMLR %P 3--24 %U https://proceedings.mlr.press/v30/Shamir13.html %V 30 %X The problem of stochastic convex optimization with bandit feedback (in the learning community) or without knowledge of gradients (in the optimization community) has received much attention in recent years, in the form of algorithms and performance upper bounds. However, much less is known about the inherent complexity of these problems, and there are few lower bounds in the literature, especially for nonlinear functions. In this paper, we investigate the attainable error/regret in the bandit and derivative-free settings, as a function of the dimension d and the available number of queries T. We provide a precise characterization of the attainable performance for strongly-convex and smooth functions, which also imply a non-trivial lower bound for more general problems. Moreover, we prove that in both the bandit and derivative-free setting, the required number of queries must scale at least quadratically with the dimension. Finally, we show that on the natural class of quadratic functions, it is possible to obtain a “fast” O(1/T) error rate in terms of T, under mild assumptions, even without having access to gradients. To the best of our knowledge, this is the first such rate in a derivative-free stochastic setting, and holds despite previous results which seem to imply the contrary.
RIS
TY - CPAPER TI - On the Complexity of Bandit and Derivative-Free Stochastic Convex Optimization AU - Ohad Shamir BT - Proceedings of the 26th Annual Conference on Learning Theory DA - 2013/06/13 ED - Shai Shalev-Shwartz ED - Ingo Steinwart ID - pmlr-v30-Shamir13 PB - PMLR DP - Proceedings of Machine Learning Research VL - 30 SP - 3 EP - 24 L1 - http://proceedings.mlr.press/v30/Shamir13.pdf UR - https://proceedings.mlr.press/v30/Shamir13.html AB - The problem of stochastic convex optimization with bandit feedback (in the learning community) or without knowledge of gradients (in the optimization community) has received much attention in recent years, in the form of algorithms and performance upper bounds. However, much less is known about the inherent complexity of these problems, and there are few lower bounds in the literature, especially for nonlinear functions. In this paper, we investigate the attainable error/regret in the bandit and derivative-free settings, as a function of the dimension d and the available number of queries T. We provide a precise characterization of the attainable performance for strongly-convex and smooth functions, which also imply a non-trivial lower bound for more general problems. Moreover, we prove that in both the bandit and derivative-free setting, the required number of queries must scale at least quadratically with the dimension. Finally, we show that on the natural class of quadratic functions, it is possible to obtain a “fast” O(1/T) error rate in terms of T, under mild assumptions, even without having access to gradients. To the best of our knowledge, this is the first such rate in a derivative-free stochastic setting, and holds despite previous results which seem to imply the contrary. ER -
APA
Shamir, O.. (2013). On the Complexity of Bandit and Derivative-Free Stochastic Convex Optimization. Proceedings of the 26th Annual Conference on Learning Theory, in Proceedings of Machine Learning Research 30:3-24 Available from https://proceedings.mlr.press/v30/Shamir13.html.

Related Material