Robust Dual View Deep Agent
This is an incremental improvement for reinforcement learning applications, specifically in domains like video game AI.
The paper tackles the problem of improving robustness and training efficiency in deep reinforcement learning agents by proposing a modified A3C architecture that splits input into two independent streams, achieving a 30% reduction in training parameters while maintaining similar playing performance.
Motivated by recent advance of machine learning using Deep Reinforcement Learning this paper proposes a modified architecture that produces more robust agents and speeds up the training process. Our architecture is based on Asynchronous Advantage Actor-Critic (A3C) algorithm where the total input dimensionality is halved by dividing the input into two independent streams. We use ViZDoom, 3D world software that is based on the classical first person shooter video game, Doom, as a test case. The experiments show that in comparison to single input agents, the proposed architecture succeeds to have the same playing performance and shows more robust behavior, achieving significant reduction in the number of training parameters of almost 30%.