Timezone: »

On Context-Dependent Clustering of Bandits
Claudio Gentile · Shuai Li · Purushottam Kar · Alexandros Karatzoglou · Giovanni Zappella · Evans Etrue Howard

Sun Aug 06 10:48 PM -- 11:06 PM (PDT) @ C4.1

We investigate a novel cluster-of-bandit algorithm CAB for collaborative recommendation tasks that implements the underlying feedback sharing mechanism by estimating user neighborhoods in a context-dependent manner. CAB makes sharp departures from the state of the art by incorporating collaborative effects into inference, as well as learning processes in a manner that seamlessly interleaves explore-exploit tradeoffs and collaborative steps. We prove regret bounds for CAB under various data-dependent assumptions which exhibit a crisp dependence on the expected number of clusters over the users, a natural measure of the statistical difficulty of the learning task. Experiments on production and real-world datasets show that CAB offers significantly increased prediction performance against a representative pool of state-of-the-art methods.

Author Information

Claudio Gentile (Universita dell'Insubria)
Shuai Li (University of Cambridge)
Purushottam Kar (Indian Institute of Technology Kanpur)
Alexandros Karatzoglou (Telefonica Research)
Giovanni Zappella (Amazon Dev Center Germany)
Evans Etrue Howard (University of Insubria)

I am currently a PhD candidate (University of Insubria, Italy) in Computational Mathematics with research interests in * On-line (sequential) prediction algorithms (e.g., Bandits) * Machine learning/Data Mining on networked data * Experimental investigations related to the above * Thompson Sampling

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors