Recent developments in autoconstructive evolution

Genetic Programming Theory and Practice VIII, 2011

Most genetic programming systems use hard-coded genetic operators that are applied according to user-specified parameters. Because it is unlikely that the provided operators or the default parameters will be ideal for all problems or all program representations, practitioners often devote considerable energy to experimentation with alternatives. Attempts to bring choices about operators and parameters under evolutionary control, through self-adaptative algorithms or meta-genetic programming, have been explored in the literature and have produced interesting results. However, no systems based on such principles have yet been demonstrated to have greater practical problem-solving power than the more-standard alternatives. This chapter explores the prospects for extending the practical power of genetic programming through the refinement of an approach called autoconstructive evolution, in which the algorithms used for the reproduction and variation of evolving programs are encoded in the programs themselves, and are thereby subject to variation and evolution in tandem with their problem-solving components. We present the motivation for the autoconstructive evolution approach, show how it can be instantiated using the Push programming language, summarize previous results with the Pushpop system, outline the more recent AutoPush system, and chart a course for future work focused on the production of practical systems that can solve hard problems.

2012

Abstract While most hyper-heuristics search for a heuristic that is later used to solve classes of problems, autoconstructive evolution represents an alternative which simultaneously searches both heuristic and solution space. In this study we contrast autoconstructive evolution, in which intergenerational variation is accomplished by the evolving programs themselves, with a genetic programming system, PushGP, to understand the dynamics of this hybrid approach.

Proceedings of the 2016 on Genetic and Evolutionary Computation Conference Companion, 2016

In autoconstructive evolutionary algorithms, individuals implement not only candidate solutions to specified computational problems, but also their own methods for variation of offspring. This makes it possible for the variation methods to themselves evolve, which could, in principle, produce a system with an enhanced capacity for adaptation and superior problem solving power. Prior work on autoconsruction has explored a range of system designs and their evolutionary dynamics, but it has not solved hard problems. Here we describe a new approach that can indeed solve at least some hard problems. We present the key components of this approach, including the use of linear genomes for hierarchically structured programs, a diversity-maintaining parent selection algorithm, and the enforcement of diversification constraints on offspring. We describe a software synthesis benchmark problem that our new approach can solve, and we present visualizations of data from single successful runs of autoconstructive vs. non-autoconstructive systems on this problem. While anecdotal, the data suggests that variation methods, and therefore significant aspects of the evolutionary process, evolve over the course of the autoconstructive runs.

1993

An evolutionary method for designing autonomous systems is proposed. The research is a computer exploration on how the global behavior of autonomous systems can emerge from neural circuits. The evolutionary approach is used to increase the repertoire of behaviors.

2002

Push is a programming language designed for the expression of evolving programs within an evolutionary computation system. This article describes Push and illustrates some of the opportunities that it presents for evolutionary computation. Two evolutionary computation systems, PushGP and Pushpop, are described in detail. PushGP is a genetic programming system that evolves Push programs to solve computational problems. Pushpop, an "autoconstructive evolution" system, also evolves Push programs but does so while simultaneously evolving its own evolutionary mechanisms.

2004

This paper describes a new Evolutionary Programming algorithm based on Self-Organised Criticality. When tested on a range of problems drawn from real-world applications in science and engineering, it performed better than a variety of gradient descent, direct search and genetic algorithms. It proved capable of delivering high quality results faster, and is simple, robust and highly parallel.

IEEE Transactions on Evolutionary Computation, 1997

Evolutionary computation has started to receive significant attention during the last decade, although the origins can be traced back to the late 1950's. This article surveys the history as well as the current state of this rapidly growing field. We describe the purpose, the general structure, and the working principles of different approaches, including genetic algorithms (GA) [with links to genetic programming (GP) and classifier systems (CS)], evolution strategies (ES), and evolutionary programming (EP) by analysis and comparison of their most important constituents (i.e., representations, variation operators, reproduction, and selection mechanism). Finally, we give a brief overview on the manifold of application domains, although this necessarily must remain incomplete.

Genetic Programming and Evolvable Machines, 2007

2016 IEEE Congress on Evolutionary Computation (CEC), 2016

Automatic Programming has long been a sub-goal of Artificial Intelligence (AI). It is feasible in limited domains. Genetic Improvement (GI) has expanded these dramatically to more than 100 000 lines of code by building on human written applications. Further scaling may need key advances in both Search Based Software Engineering (SBSE) and Evolutionary Computation (EC) research, particularly on representations, genetic operations, fitness landscapes, fitness surrogates, multi objective search and co-evolution.

… , 2001. Proceedings of …, 2001

In the last five years, the field of evolutionary computation (EC) has seen a resurgence of new ideas, many stemming from new biological inspirations. This paper outlines four of these new branches of research: Creative Evolutionary Systems, Computational Embryology, Evolvable Hardware and Artificial Immune Systems, showing how they aim to extend the capabilities of EC. Recent, unpublished results by researchers in each area at the Department of Computer Science, UCL are provided.