Evolutionary Learning: Advances in Theories and Algorithms

Abstract

of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Near-Optimal Graph Signal Sampling by Pareto Optimization

“…The proof of Theorem 1 relies on Lemma 2 and 3 stated bellow. The complete proof is similar to the proof of theorem 14.1 in [ 22 ], which is omitted due to the limited space.…”

Evolutionary Learning for Soft Margin Problems: A Case Study on Practical Problems with Kernels

“…Evolutionary strategy is a bio-inspired method for search and optimisation problems, and it mimics the natural environments, criteria and processes. There are many well-used evolutionary operators, such as simulated binary crossover (SBX) [22], polynomial-based mutation (PM) [23] and others [11,13]. Here we take SBX as an example to introduce evolutionary strategy.…”

“…For the second issue (Issue B), evolutionary learning [13] works, and it is a type of bio-inspired method for solving highly complex, non-linear and larger-scale optimization problems by combining an evolutionary strategy with machine learning [7,[13][14][15][16][17]. Compared to the traditional gradient methods, evolutionary learning has shown advantages in tackling practical problems, especially those which are nondifferentiable or very difficult to model mathematically.…”

HyperTube: A Framework for Population-Based Online Hyperparameter Optimization with Resource Constraints

“…To improve this approach, further exploration and exploitation techniques, such as evolutionary algorithm and Bayesian optimization, could be introduced. For example, by using a genetic algorithm (GA), which includes mutation and crossover procedures, we can improve the population by adding the children of selected configurations with good performance [37], [38]. The crossover procedure is performed parameter-wise, which means we randomly exchange the whole value of each hyperparameter within the list of hyperparameters to be searched.…”