The evolution of learning systems:to Bayes or not to be

Bayesian algorithms pose a limit to the performance learning algorithms can achieve. Natural selection should guide the evolution of information processing systems towards those limits. What can we learn from this evolution and what properties do the intermediate stages have? While this question is too general to permit any answer, progress can be made by restricting the class of information processing systems under study. We present analytical and numerical results for the evolution of on-line algorithms for learning from examples for neural network classifiers, which might include or not a hidden layer. The analytical results are obtained by solving a variational problem to determine the learning algorithm that leads to maximum generalization ability. Simulations using evolutionary programming, for programs that implement learning algorithms, confirm and expand the results. The principal result is not just that the evolution is towards a Bayesian limit. Indeed it is essentially reached. In addition we find that evolution is driven by the discovery of useful structures or combinations of variables and operators. In different runs the temporal order of the discovery of such combinations is unique. The main result is that combinations that signal the surprise brought by an example arise always before combinations that serve to gauge the performance of the learning algorithm. This latter structures can be used to implement annealing schedules. The temporal ordering can be understood analytically as well by doing the functional optimization in restricted functional spaces. We also show that there is data suggesting that the appearance of these traits also follows the same temporal ordering in biological systems. © 2006 American Institute of Physics.

Evolutionary neural networks : models and applications

The scaling problems which afflict attempts to optimise neural networks (NNs) with genetic algorithms (GAs) are disclosed. A novel GA-NN hybrid is introduced, based on the bumptree, a little-used connectionist model. As well as being computationally efficient, the bumptree is shown to be more amenable to genetic coding lthan other NN models. A hierarchical genetic coding scheme is developed for the bumptree and shown to have low redundancy, as well as being complete and closed with respect to the search space. When applied to optimising bumptree architectures for classification problems the GA discovers bumptrees which significantly out-perform those constructed using a standard algorithm. The fields of artificial life, control and robotics are identified as likely application areas for the evolutionary optimisation of NNs. An artificial life case-study is presented and discussed. Experiments are reported which show that the GA-bumptree is able to learn simulated pole balancing and car parking tasks using only limited environmental feedback. A simple modification of the fitness function allows the GA-bumptree to learn mappings which are multi-modal, such as robot arm inverse kinematics. The dynamics of the 'geographic speciation' selection model used by the GA-bumptree are investigated empirically and the convergence profile is introduced as an analytical tool. The relationships between the rate of genetic convergence and the phenomena of speciation, genetic drift and punctuated equilibrium arc discussed. The importance of genetic linkage to GA design is discussed and two new recombination operators arc introduced. The first, linkage mapped crossover (LMX) is shown to be a generalisation of existing crossover operators. LMX provides a new framework for incorporating prior knowledge into GAs.Its adaptive form, ALMX, is shown to be able to infer linkage relationships automatically during genetic search.

Towards identifying salient patterns in genetic programming individuals

This thesis addresses the problem of offline identification of salient patterns in genetic programming individuals. It discusses the main issues related to automatic pattern identification systems, namely that these (a) should help in understanding the final solutions of the evolutionary run, (b) should give insight into the course of evolution and (c) should be helpful in optimizing future runs. Moreover, it proposes an algorithm, Extended Pattern Growing Algorithm ([E]PGA) to extract, filter and sort the identified patterns so that these fulfill as many as possible of the following criteria: (a) they are representative for the evolutionary run and/or search space, (b) they are human-friendly and (c) their numbers are within reasonable limits. The results are demonstrated on six problems from different domains.

Evolution of lace knitting stitch patterns by genetic programming

In this paper we study the generation of lace knitting stitch patterns by using genetic programming. We devise a genetic representation of knitting charts that accurately reflects their usage for hand knitting the pattern. We apply a basic evolutionary algorithm for generating the patterns, where the key of success is evaluation. We propose automatic evaluation of the patterns, without interaction with the user. We present some patterns generated by the method and then discuss further possibilities for bringing automatic evaluation closer to human evaluation. Copyright 2007 ACM.

Two-B or not two-B? Design patterns for hybrid metaheuristics

Real world search problems, characterised by nonlinearity, noise and multidimensionality, are often best solved by hybrid algorithms. Techniques embodying different necessary features are triggered at specific iterations, in response to the current state of the problem space. In the existing literature, this alternation is managed either statically (through pre-programmed policies) or dynamically, at the cost of high coupling with algorithm inner representation. We extract two design patterns for hybrid metaheuristic search algorithms, the All-Seeing Eye and the Commentator patterns, which we argue should be replaced by the more flexible and loosely coupled Simple Black Box (Two-B) and Utility-based Black Box (Three-B) patterns that we propose here. We recommend the Two-B pattern for purely fitness based hybridisations and the Three-B pattern for more generic search quality evaluation based hybridisations.