Partner-Aware Algorithms in Decentralized Cooperative Bandit Teams

Authors: Erdem Biyik, Anusha Lalitha, Rajarshi Saha, Andrea Goldsmith, Dorsa Sadigh9296-9303

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We analytically show that our proposed strategy achieves logarithmic regret, and provide extensive experiments involving human-AI and human-robot collaboration to validate our theoretical findings.
Researcher Affiliation Academia 1 Department of Electrical Engineering, Stanford University 2 Department of Electrical and Computer Engineering, Princeton University 3 Department of Computer Science, Stanford University
Pseudocode Yes Algorithm 1: Partner-Aware UCB: Follower; Algorithm 2: Partner-Aware UCB: Leader
Open Source Code Yes Code at: https://sites.google.com/view/partner-aware-ucb
Open Datasets No The paper describes conducting simulations with fixed or random reward means and human-subject studies using an experimental setup (burger stacking, slot machines), but does not use a traditional publicly available or open dataset with access information.
Dataset Splits No The paper describes experimental runs and user studies, including warm-starting with simulated agents, but it does not specify traditional training/validation/test dataset splits needed for reproduction in a typical ML context.
Hardware Specification No The paper mentions using a 'Fetch robot (Wise et al. 2016)' for human-subject studies but does not provide specific hardware specifications (e.g., CPU, GPU, memory) of the computational resources used for simulations or training models, nor the robot's internal processing hardware.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies, libraries, or programming languages used in the experiments or simulations.
Experiment Setup Yes Unless otherwise noted |A1|=|A2|=2, p1 =1, p2 =0.5, c(L) =c(F) =0.025 in these simulations. ... For Partner Aware UCB, we set L = 1, W = 25. ... We set, when relevant, L = 1, W = 2 and c(L) = c(F) = 0.01. ... we set c(i) = 0.025, L = 1, and W = 25 for all agents.