Non-Asymptotic Analysis for Single-Loop (Natural) Actor-Critic with Compatible Function Approximation

Authors: Yudan Wang, Yue Wang, Yi Zhou, Shaofeng Zou

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical results are also provided in the appendix. In this section, we conduct experiments to numerically verify our AC/NAC with compatible function approximation.
Researcher Affiliation Academia 1Electrical Engineering, University at Buffalo 2Electrical and Computer Engineering, University of Central Florida 3Electrical and Computer Engineering, University of Utah 4Computer Science & Engineering, University at Buffalo.
Pseudocode Yes Algorithm 1 (Natural) Actor-Critic with Compatible Function Approximation; Algorithm 2 Compatible k-step TD Algorithm
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes We test our algorithms in the Acrobot environment (Sutton, 1995).
Dataset Splits No The paper uses the Acrobot environment but does not specify explicit training, validation, or test dataset splits.
Hardware Specification No The paper does not provide specific hardware details used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers.
Experiment Setup Yes In our experiment setup, we set k = 128, and design a 2-layer neural network with 16 hidden neurons to represent the policy, which contains 163 parameters.