ABSTRACT
In this paper, we address the problem of black box optimization over binary vectors. We introduce a novel evolutionary algorithm called Herding Evolutionary Algorithm which relies on herding to generate individuals with empirical moments close to those of selected individuals. We report experiments with diverse fitness functions and compare the results of HEA with those of simulated annealing, local search, and PBIL. Although in its early developments, HEA shows promising results.
- F. Bach, S. Lacoste-Julien, and G. Obozinski. On the equivalence between herding and conditional gradient algorithms. In International Conference on Machine Learning, 2012.Google ScholarDigital Library
- S. Baluja and R. Caruana. Removing the genetics from the standard genetic algorithm. In Proceedings of the 12th Annual Conference on Machine Learning, pages 38--46, 1995.Google ScholarCross Ref
- S. Baluja and S. Davies. Using optimal dependency-trees for combinatorial optimization: learning the structure of the search space. Technical Report Carnegie Mellon University-CS-97-107, Carnegie-Mellon University, January 1997.Google Scholar
- Y. Chen, M. Welling, and A. Smola. Super-samples from kernel herding. In Uncertainty in Artificial Intelligence, pages 109--116, 2010.Google Scholar
- M. Sebag and M. Schoenauer. A society of hill-climbers. In Proc. IEEE Int. Conf. on Evolutionary Computation, pages 319--324, Indianapolis, April 1997.Google ScholarCross Ref
- M. Welling. Herding dynamic weights for partially observed random field models. In Uncertainty in Artificial Intelligence, pages 599--606, 2009. Google ScholarDigital Library
- M. Welling. Herding dynamic weights to learn. In International Conference on Machine Learning, pages 1121--1128, 2009. Google ScholarDigital Library
Index Terms
- Herding Evolutionary Algorithm
Recommendations
Benchmarking parameter-free amalgam on functions with and without noise
We describe a parameter-free estimation-of-distribution algorithm EDA called the adapted maximum-likelihood Gaussian model iterated density-estimation evolutionary algorithm AMaLGaM-ID<inline-formula><inline-graphic xlink="EVCO_a_00094inline1.gif" ...
Double elite co-evolutionary genetic algorithm
A new double elite co-evolutionary genetic algorithm is proposed to avoid the premature convergence and low speed of convergence based on the elite strategy and the concept of co-evolution. In the DECGA, the two different and high fitness individuals (...
An Endosymbiotic Evolutionary Algorithm for Optimization
This paper proposes a new symbiotic evolutionary algorithm to solve complex optimization problems. This algorithm imitates the natural evolution process of endosymbionts, which is called endosymbiotic evolutionary algorithm. Existing symbiotic ...
Comments