Kernel Methods for Cooperative Multi-Agent Contextual Bandits

Authors: Abhimanyu Dubey, Alex ‘Sandy’ Pentland

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, on a series of both synthetic and real-world multi-agent network benchmarks, we demonstrate that our algorithm significantly outperforms existing benchmarks.
Researcher Affiliation Academia 1Media Lab and Institute for Data, Systems and Society, Massachusetts Institute of Technology. Correspondence to: Abhimanyu Dubey <dubeya@mit.edu>.
Pseudocode Yes Algorithm 1 COOP-KERNELUCB
Open Source Code No The paper does not provide a link to open-source code or explicitly state that the code is publicly available.
Open Datasets Yes For the synthetic networks, we subsample V nodes and their corresponding edges (for V = 200) from the ego-Facebook, musae-Twitch, and as-Skitter networks, in order to represent a diverse set of networks found in social networks, peer-to-peer distribution and autonomous systems.
Dataset Splits No The paper mentions generating contexts and running experiments, but does not specify explicit training, validation, or test dataset splits in terms of percentages or sample counts. It states: "Dv,t is a set of 8 randomly generated contexts for all v V, t T and dimensionality d = 10 for X and Z (for setup 2)."
Hardware Specification No The paper does not provide specific details about the hardware used for running the experiments (e.g., GPU/CPU models, memory, cloud instances).
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes For the kernel estimation task, we set σ = 1, and we set λ = 1.