[Back]

M. Lechner, R. Hasani, R. Grosu:
"Interpretable Neuronal Circuit Policies for Reinforcement Learning Environments";
Talk: Workshop on Explainable Artificial Intelligence (XAI2018) at IJCAI-ECAI 2018, Stockholm, Sweden; 2018-07-13 - 2018-07-18; in: "Proceedings of the 2nd Workshop on Explainable Artificial Intelligence", IJCAI-ECAI 2018, (2018), 79 - 84.

We propose an effective way to create interpretable control agents, by re-purposing the function of a biological neural circuit model, to govern simulated and real world reinforcement learning (RL) test-beds.
We model the tap-withdrawal (TW) neural circuit of the nematode, C. elegans, a circuit responsible for the worm´s reflexive response to external mechanical touch stimulations, and learn its synaptic
and neuronal parameters as a policy for controlling basic RL tasks. We also autonomously park a real rover robot on a pre-defined trajectory, by deploying such neuronal circuit policies learned in a simulated
environment. For reconfiguration of the purpose of the TW neural circuit, we adopt a search-based RL algorithm. We show that our neuronal policies perform as good as deep neural network policies with
the advantage of realizing interpretable dynamics at the cell level.