Online Structured Prediction via Coactive Learning

TL;DR

This paper introduces Coactive Learning, a framework where systems learn from user feedback, such as clicks, to improve predictions like search rankings, achieving low regret without explicit utility scores.

Contribution

It presents a novel Coactive Learning model that infers utility from observable user behavior and develops algorithms with provable regret bounds for structured prediction tasks.

Findings

Algorithms achieve ${\cal O}(\frac{1}{\sqrt{T}})$ regret

Effective in web-search ranking tasks

Applicable to movie recommendation systems

Abstract

We propose Coactive Learning as a model of interaction between a learning system and a human user, where both have the common goal of providing results of maximum utility to the user. At each step, the system (e.g. search engine) receives a context (e.g. query) and predicts an object (e.g. ranking). The user responds by correcting the system if necessary, providing a slightly improved -- but not necessarily optimal -- object as feedback. We argue that such feedback can often be inferred from observable user behavior, for example, from clicks in web-search. Evaluating predictions by their cardinal utility to the user, we propose efficient learning algorithms that have ${\cal O}(\frac{1}{\sqrt{T}})$ average regret, even though the learning algorithm never observes cardinal utility values as in conventional online learning. We demonstrate the applicability of our model and learning algorithms on a movie recommendation task, as well as ranking for web-search.