The CMA Evolution Strategy: A Tutorial
About
This tutorial introduces the CMA Evolution Strategy (ES), where CMA stands for Covariance Matrix Adaptation. The CMA-ES is a stochastic, or randomized, method for real-parameter (continuous domain) optimization of non-linear, non-convex functions. We try to motivate and derive the algorithm from intuitive concepts and from requirements of non-linear, non-convex search in continuous domain.
Nikolaus Hansen• 2016
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Image Generation | Imagenet-256 latent space | FID6.99 | 102 | |
| Multi-turn Conversation Evaluation | MT-Bench | MT-Bench Score7.43 | 68 | |
| Circuit Optimization | FDDSD Gm circuit | Figure of Merit (FoM)7.23 | 25 | |
| Circuit Optimization | Bandgap circuit | FoM6.28 | 25 | |
| Black-box Optimization | BBOB d=100 | F41.68e+3 | 25 | |
| Circuit Optimization | Three-stage circuit | FoM6.76 | 25 | |
| Circuit Optimization | Two-stage circuit | FoM6.33 | 25 | |
| Circuit Optimization | Charge Pump circuit | FoM18.05 | 25 | |
| Circuit Optimization | LDO circuit | FoM10.0217 | 25 | |
| Preference Prediction | RewardBench | Accuracy74 | 25 |
Showing 10 of 105 rows
...