Structured Evolution with Compact Architectures for Scalable Policy Optimization
This work addresses scalable policy optimization for resource-constrained real hardware, such as robotics, by providing more efficient training and inference through compact architectures.
The paper tackles the problem of blackbox optimization for learning compact policies by introducing a method using structured random orthogonal matrices for gradient approximation, achieving up to 13x fewer parameters than Evolution Strategies and enabling neural networks with less than 300 parameters to solve most OpenAI Gym robotics tasks with almost linear inference time.
We present a new method of blackbox optimization via gradient approximation with the use of structured random orthogonal matrices, providing more accurate estimators than baselines and with provable theoretical guarantees. We show that this algorithm can be successfully applied to learn better quality compact policies than those using standard gradient estimation techniques. The compact policies we learn have several advantages over unstructured ones, including faster training algorithms and faster inference. These benefits are important when the policy is deployed on real hardware with limited resources. Further, compact policies provide more scalable architectures for derivative-free optimization (DFO) in high-dimensional spaces. We show that most robotics tasks from the OpenAI Gym can be solved using neural networks with less than 300 parameters, with almost linear time complexity of the inference phase, with up to 13x fewer parameters relative to the Evolution Strategies (ES) algorithm introduced by Salimans et al. (2017). We do not need heuristics such as fitness shaping to learn good quality policies, resulting in a simple and theoretically motivated training mechanism.