CoNES: Convex Natural Evolutionary Strategies
arXiv:2007.08601
Abstract
We present a novel algorithm -- convex natural evolutionary strategies (CoNES) -- for optimizing high-dimensional blackbox functions by leveraging tools from convex optimization and information geometry. CoNES is formulated as an efficiently-solvable convex program that adapts the evolutionary strategies (ES) gradient estimate to promote rapid convergence. The resulting algorithm is invariant to the parameterization of the belief distribution. Our numerical results demonstrate that CoNES vastly outperforms conventional blackbox optimization methods on a suite of functions used for benchmarking blackbox optimizers. Furthermore, CoNES demonstrates the ability to converge faster than conventional blackbox methods on a selection of OpenAI's MuJoCo reinforcement learning tasks for locomotion.
References in corpus (7)
- Adam: A Method for Stochastic Optimization
- Evolution Strategies as a Scalable Alternative to Reinforcement Learning
- Information-Geometric Optimization Algorithms: A Unifying Picture via Invariance Principles
- Simple random search provides a competitive approach to reinforcement learning
- Fisher-Rao Metric, Geometry, and Complexity of Neural Networks
- Mirror Descent Search and its Acceleration
- Provably Robust Blackbox Optimization for Reinforcement Learning