Constrained Cross-Entropy Method for Safe Reinforcement Learning

Authors: Min Wen, Ufuk Topcu

NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We show with simulation experiments that the proposed algorithm can effectively learn feasible policies without assumptions on the feasibility of initial policies, even with non-Markovian objective functions and constraint functions.
Researcher Affiliation Academia Min Wen Department of Electrical and Systems Engineering University of Pennsylvania wenm@seas.upenn.edu
Pseudocode Yes Algorithm 1 Constrained Cross-Entropy Method
Open Source Code No The paper does not provide an explicit statement or link for the open-source code of the described methodology.
Open Datasets No The paper describes a 'mobile robot navigation task' in a simulated environment but does not mention using a publicly available dataset, nor does it provide concrete access information (link, DOI, citation) for any data used.
Dataset Splits No The paper does not provide specific details on training, validation, or test dataset splits. It discusses evaluation based on 'learning curves' but not data partitioning.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory) used for running its experiments. It only mentions implementation in 'rllab [5]'.
Software Dependencies No The paper states 'All experiments are implemented in rllab [5]' but does not provide version numbers for rllab or any other software dependencies.
Experiment Setup Yes For all experiments, the agent s policy is modeled as a fully connected neural network with two hidden layers with 30 nodes in each layer. Trajectory length for all experiments is set to N = 30.