Adapting arbitrary normal mutation distributions in evolution strategies: the covariance matrix adaptation

By Nikolaus Hansen, Andreas Ostermeier. 1996

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Optimization & Training

Citations in the corpus

Trust Region Policy Optimization (2015, methodology): “…TRPO ; cross-entropy method (CEM), a gradient-free method (Szita & L¨orincz, 2006); covariance matrix adaption (CMA), another gradient-free method (Hansen & Ostermeier, 1996); natural gradient , the classic natural policy gradient algorithm (Kakade, 2002), which differs from single path by the…”
One Pixel Attack for Fooling Deep Neural Networks (2017, methodology): “arch belongs to a big class of algorithms called evolutionary strategies [48] which includes other variants such as Adaptive DE [49] and Covariance matrix adaptation evolution strategy (CMA-ES) [50], [51], [52]. In fact, there are a couple of recent developments [61], [62], [63] in evolutionary stra”
Automated Machine Learning on Big Data using Stochastic Algorithm Tuning (2014, methodology): “We also tested covariance matrix adaptive evolution strategy (CMAES) [25] and dividing rectangles (DIRECT) [26] however found that these did not converge on this very noisy objective.”
Evaluating the CMA Evolution Strategy on Multimodal Test Functions (2004, methodology): “2 Experimental procedure The performance of the CMA-ES is tested for dimensions n = [2, 5, 10, 20, 40, 80].”
Evolution strategies – A comprehensive introduction (2002, unknown): “Therefore, the expected v will be smaller than u and σ should be decreased (see also Hansen and Ostermeier, 1996).”
Evolutionary algorithms for the optimization of advective control of contaminated aquifer zones (2004, methodology): “A main feature of the DES is the covariance matrix adaptation (CMA) introduced by Hansen and Ostermeier [1996] to achieve a rotation invariant implementation of both individual and global step sizes.”
Self-Adaptive Multimethod Search for Global Optimization in Real-Parameter Spaces (2009, background): “strategy that adapts a full covariance matrix of a normal search (mutation) distribution [18]–[20].”
Optimal gait and form for animal locomotion (2009, methodology): “…contact times and aid in avoiding local minima, we employ a hybrid optimization technique which combines a derivative-aware spacetime constraints approach as the inner loop to a derivative-free global optimization based on the covariance matrix adaptation evolution strategy [Hansen et al. 1996].”
A restart CMA evolution strategy with increasing population size (2005, methodology): “The Covariance Matrix Adaptation Evolution Strategy (CMA-ES) [5, 7, 4] is an evolution strategy that adapts the full covariance matrix of a normal search (mutation) distribution.”
The CMA Evolution Strategy: A Comparing Review (2006, methodology): “The CMA Evolution Strategy: A Comparing Review C (g+1) λ = 1 λ λ i=1 x (g+1) i − m (g) x (g+1) i − m (g) T (7) The matrix C (g+1) λ is an unbiased maximum likelihood estimator of C (g) .”
A comparative study of Artificial Bee Colony algorithm (2009, methodology): “In Covariance Matrix Adaptation Evolution Strategies (CMA-ES) devised for optimal exploitation of local information, the step size can be selected by self-adaptation or by covariance matrix adaptation [40].”
Multiobjective optimization using evolutionary algorithms (2000, methodology): “It was first described by Hansen & Ostermeier (1996) and further improved and evaluated by Hansen & Ostermeier (1997).”
Evolution Strategies (2018, methodology): “The covariance matrix adaptation evolution strategy (CMA-ES) [27, 10, 36] is a de facto standard in continuous domain evolutionary computation.”
Ray RLLib: A Composable and Scalable Reinforcement Learning Library (2017, unknown): “] X X DQNs [38, 49, 57, 59, 14, 29, 41] X X X Off-policy PG [35, 25, 58, 30] X X X Model-Based/Hybrid [60, 45, 42] X X Model-Based Planning Multi-Agent [36, 20, 39, 46, 21] X X X Evolutionary Methods [47, 27] X Derivative-Free Optimization AlphaGo [53, 54] X X X MCTS, Derivative-Free Optimization 2 ”
Parameter Control in Evolutionary Algorithms: Trends and Challenges (2015, methodology): “The concept of an efficient evolution path motivated the use of correlation between individual mutation step sizes of the dimensions of the problem and the adaptation of a covariance matrix [92], [158].”