Deep Reinforcement Learning for Optimization at Early Design Stages
Lorenzo Servadei, Jin Hwa Lee, José A Arjona Medina, Michael Werner, Sepp Hochreiter, Wolfgang Ecker, and Robert Wille
In this paper, we introduce Deep Reinforcement Learning (DRL) for design cost optimization at early stages of the System on Chips (SoCs) design process. We demonstrate that DRL is a suitable solution for the problem at hand. We benchmark three DRL algorithms based on Pointer Network, a neural network specifically applied for combinatorial problems, on the design cost optimization. We show that this lead to the considerable improvements in cost optimization compared to conventional optimization methods. Additionally, by using the recently introduced RUDDER method and its reward redistribution approach, we obtain a significant improvement in complex designs.
IEEE Design & Test, 2022-01-20.