Collaborative Multi-Agent Bandits

Our OPODIS 2023 paper (Flooding with Absorption: An Efficient Protocol for Heterogeneous Bandits over Complex Networks) has received the best student paper award! Again, this is joint work with Laura Schmid and Se-Young Yun (KAIST AI).

Our preprint Improved Regret Bounds of (Multinomial) Logistic Bandits via Regret-to-Confidence-Set Conversion is now available on arXiv! (https://arxiv.org/abs/2310.18554) This is joint work with Kwang-Sung Jun (Univ. of Arizona CS) and Se-Young Yun (KAIST AI).

One paper (Flooding with Absorption: An Efficient Protocol for Heterogeneous Bandits over Complex Networks) is accepted to OPODIS 2023! This is joint work with Laura Schmid and Se-Young Yun (KAIST AI).

We have updated our arXiv paper on collaborative bandits over a network, with the new title Flooding with Absorption: An Efficient Protocol for Heterogeneous Bandits over Complex Networks as well as several significant changes in the overall organization and experiments.

Our preprint Communication-Efficient Collaborative Heterogeneous Bandits in Networks is now available on arXiv! (https://arxiv.org/abs/2303.05445) This is joint work with Laura Schmid and Se-Young Yun (KAIST AI). This is the full version of our MobiHoc 2023 submission of the same title.

A novel problem setting where heterogeneous multi-agent bandits collaborate over a network to minimize their group regret. To deal with the high communication complexity of the classic flooding protocol combined with UCB, a new network protocol called Flooding with Absorption (FwA) is proposed. Theoretical and empirical analyses are provded for flooding and FwA, showing the efficacy of our proposed FwA.