Safety-Oriented Pruning and Interpretation of Reinforcement Learning Policies

Dennis Gross; Helge Spieker

Safety-Oriented Pruning and Interpretation of Reinforcement Learning Policies

Machine Learning 2024-09-17 v1

Authors: Dennis Gross , Helge Spieker

Abstract

Pruning neural networks (NNs) can streamline them but risks removing vital parameters from safe reinforcement learning (RL) policies. We introduce an interpretable RL method called VERINTER, which combines NN pruning with model checking to ensure interpretable RL safety. VERINTER exactly quantifies the effects of pruning and the impact of neural connections on complex safety properties by analyzing changes in safety measurements. This method maintains safety in pruned RL policies and enhances understanding of their safety dynamics, which has proven effective in multiple RL settings.

Keywords

safe reinforcement learning neural network pruning interpretable reinforcement learning model checking

Cite

@article{arxiv.2409.10218,
  title  = {Safety-Oriented Pruning and Interpretation of Reinforcement Learning Policies},
  author = {Dennis Gross and Helge Spieker},
  journal= {arXiv preprint arXiv:2409.10218},
  year   = {2024}
}

Related papers

View all related →