Evolutionary Algorithms for Neural Network Design and Training

By Jürgen Branke. 1995

How researchers see this paper

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

Citations in the corpus

Efficient Progressive Neural Architecture Search (2018, background): “Researchers quickly realized that, due to the extremely high computational demands of the problem, neural topology optimization would need to wait for the advent of higher-order parallel computing hardware and software [3].”
A Hybrid Neurogenetic Approach for Stock Forecasting (2007, background): “Various schemes for combining GAs and NNs were proposed or compared, which were mostly designed to optimize the networks weights or to find a good topology (see survey papers [11] and [12]).”
Trends in extreme learning machines: A review (2015, methodology): “…various ways to improve the efficiency or optimality in training FNN, such as second order optimization methods (Hagan & Menhaj, 1994; Wilamowski & Yu, 2010), subset selection methods (Chen, Cowan, &Grant, 1991; Li, Peng, & Irwin, 2005) or global optimizationmethods (Branke, 1995; Yao, 1993).”
Let a biogeography-based optimizer train your Multi-Layer Perceptron (2014, background): “The literature shows that heuristic optimization methods are promising alternatives for gradient-based learning algorithms [8,78] since the stochastic nature of these algorithms allows them to avoid local minima better than gradient-based techniques and optimize challenging problems [44].”
Evolving Neural Networks through Augmenting Topologies (2002, background): “There has also been a great deal of interest in evolving network topologies as well as weights over the last decade (Angeline et al., 1993; Branke, 1995; Gruau et al., 1996; Yao, 1999).”
Global optimization issues in deep network regression: an overview (2018, methodology): “We mention more specific for FNN: simulated annealing [4,91,115]; evolutionary algorithms [28,62,97,104]; genetic algorithms [51,82,122]; particle swarm optimization (PSO) [146]; dynamic tunneling in the weight space as in [117] or in [38] where the TRUST algorithm [10,37] is applied in connection w”