Deep Radial-Basis Value Functions for Continuous Control

Authors: Kavosh Asadi, Neev Parikh, Ronald E. Parr, George D. Konidaris, Michael L. Littman6696-6704

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we empirically assess the effectiveness of RBVFs in the context of non-linear regression, valuefunction-only deep RL, and actor-critic deep RL.
Researcher Affiliation Collaboration 1 Amazon Web Services 2 Brown University 3 Duke University
Pseudocode Yes Algorithm 1 Pseudo-code for RBF-DQN
Open Source Code Yes We have released our code: 3.github.com/kavosh8/RBFDQN pytorch
Open Datasets Yes We evaluate RBF-DQN against state-of-the-art value-function-only deep RL baselines... on 6 Open AI Gym environments.
Dataset Splits No The paper uses continuous interaction with RL environments (Open AI Gym) rather than fixed static datasets with explicit train/validation/test splits. For the regression task, it mentions sampling 500 inputs for training but no explicit splits.
Hardware Specification No The paper does not explicitly describe the specific hardware (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper mentions 'pytorch' in the code release link but does not specify a version number for it or any other key software dependencies.
Experiment Setup Yes Details of our hyper-parameter tuning can be found in the Appendix.