Alex Slivkins: publications by topic

[home | by year]
All pdf links are to "full versions".

Exploration, Exploitation and Incentives. We study various scenarios in which the algorithmic challenges of online learning, and particularly the exploration-exploitation tradeoff, are intertwined with the mechanism design challenges of interacting with self-interested agents. Explore-exploit learning with resource constraints: e.g. dynamic pricing with limited supply, dynamic procurement on a budget, pay-per-click ad allocation with advertisers' budgets, etc. Crowdsourcing systems: design of algorithms and incentives. Multi-armed bandits with a similarity structure on arms and/or contexts. Multi-armed bandits in a changing environment. Contextual bandits with policy sets: algorithms and systems Social networks Algorithms for Internet and P2P networks: network triangulation and network/metric embeddings, locality-aware distributed data structures, decentralized failure detection, etc. Peer-to-peer systems