Theoretically Principled Deep RL Acceleration via Nearest Neighbor Function Approximation

Authors: Junhong Shen, Lin F. Yang9558-9566

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on classical control and Mu Jo Co locomotion tasks show that the NN-accelerated agents achieve higher sample efficiency and stability than the baseline agents.
Researcher Affiliation Academia Junhong Shen, Lin F. Yang University of California, Los Angeles jhshen@ucla.edu, linyang@ee.ucla.edu
Pseudocode Yes Algorithm 1 Nearest Neighbor Actor-Critic and Algorithm 2 Soft Nearest Neighbor Update
Open Source Code No No explicit statement or link providing access to the authors' source code for the methodology described in this paper was found.
Open Datasets Yes We use the Open AI Gym implementation (Brockman et al. 2016).
Dataset Splits No No explicit statement about specific training/validation/test dataset splits was found. The paper focuses on online reinforcement learning, where data is generated through interaction with the environment rather than pre-split datasets.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, processor types, memory amounts) used for running experiments were mentioned.
Software Dependencies No The paper mentions 'Open AI Gym implementation' and 'Stable Baselines implementation (Hill et al. 2018)' but does not provide specific version numbers for software dependencies.
Experiment Setup Yes The discount factor γ is 0.99. The Lipschitz L is determined by a grid search and set to 7. All agents are trained with 5 random seeds. Evaluation is done every 1000 steps without exploration.