Learning from eXtreme Bandit Feedback

TL;DR

This paper introduces a novel importance sampling estimator and a new algorithm, POXM, to effectively learn from bandit feedback in extremely large action spaces, significantly improving over existing methods in recommendation system tasks.

Contribution

The paper proposes the sIS estimator and the POXM algorithm, which together enable more effective learning from bandit feedback in large-scale recommendation systems, reducing bias and variance issues.

Findings

POXM outperforms baseline methods on XMC datasets.

The sIS estimator reduces variance in importance sampling.

POXM significantly improves reward estimation accuracy.

Abstract

We study the problem of batch learning from bandit feedback in the setting of extremely large action spaces. Learning from extreme bandit feedback is ubiquitous in recommendation systems, in which billions of decisions are made over sets consisting of millions of choices in a single day, yielding massive observational data. In these large-scale real-world applications, supervised learning frameworks such as eXtreme Multi-label Classification (XMC) are widely used despite the fact that they incur significant biases due to the mismatch between bandit feedback and supervised labels. Such biases can be mitigated by importance sampling techniques, but these techniques suffer from impractical variance when dealing with a large number of actions. In this paper, we introduce a selective importance sampling estimator (sIS) that operates in a significantly more favorable bias-variance regime. The sIS estimator is obtained by performing importance sampling on the conditional expectation of the reward with respect to a small subset of actions for each instance (a form of Rao-Blackwellization). We employ this estimator in a novel algorithmic procedure -- named Policy Optimization for eXtreme Models (POXM) -- for learning from bandit feedback on XMC tasks. In POXM, the selected actions for the sIS estimator are the top-p actions of the logging policy, where p is adjusted from the data and is significantly smaller than the size of the action space. We use a supervised-to-bandit conversion on three XMC datasets to benchmark our POXM method against three competing methods: BanditNet, a previously applied partial matching pruning strategy, and a supervised learning baseline. Whereas BanditNet sometimes improves marginally over the logging policy, our experiments show that POXM systematically and significantly improves over all baselines.