Symbiotic coevolutionary genetic programming: a benchmarking study under large attribute spaces

Abstract

Classification under large attribute spaces represents a dual learning problem in which attribute subspaces need to be identified at the same time as the classifier design is established. Embedded as opposed to filter or wrapper methodologies address both tasks simultaneously. The motivation for this work stems from the observation that team based approaches to Genetic Programming (GP) have the potential to design multiple classifiers per class-each with a potentially unique attribute subspace-without recourse to filter or wrapper style preprocessing steps. Specifically, competitive coevolution provides the basis for scaling the algorithm to data sets with large instance counts; whereas cooperative coevolution provides a framework for problem decomposition under a bid-based model for establishing program context. Symbiosis is used to separate the tasks of team/ ensemble composition from the design of specific team members. Team composition is specified in terms of a combinatorial search performed by a Genetic Algorithm (GA); whereas the properties of individual team members and therefore subspace identification is established under an independent GP population. Teaming implies that the members of the resulting ensemble of classifiers should have explicitly non-overlapping behaviour. Performance evaluation is conducted over data sets taken from the UCI repository with 649-102,660 attributes and 2-10 classes. The resulting teams identify attribute spaces 1-4 orders of magnitude smaller than under the original data set. Moreover, team members generally consist of less than 10 instructions; thus, small attribute subspaces are not being traded for opaque models.

Optimizing human action recognition based on a cooperative coevolutionary algorithm

“…A cooperative coevolutionary algorithm is successfully employed in Derrac et al (2010) on datasets of different data nature. McIntyre and Heywood (2011); Doucette et al (2012) combined competitive and symbiotic (cooperative) coevolution multiobjective optimisation and GP classifiers. Competition provides a mechanism for scaling to potentially large unbalanced datasets while cooperation allows the decomposition of the training set to improve the results.…”

Malicious Automatically Generated Domain Name Detection Using Stateful-SBB

“…Finally, each learner resets its registers before bidding on the next point. A more detailed explanation of the algorithm can be found in [4].…”

“…To detect these anomalies, we employ an evolutionary computation technique based on SBB [4]. Our proposed system employ a modified version of SBB, hereafter called Stateful-SBB (Stateful Symbiotic Bid-Based Genetic Programming).…”

GP under streaming data constraints

“…Possible alternatives might assume sampling under probabilistic frameworks [7], error variance sampling [8] or host-parasite interactions [2]. In the following we summarize the process adopted by the Symbiotic bid-based (SBB) framework for cooperatively evolving GP teams [11,5]. A two stage process is therefore assumed consisting of Pareto dominance ranking and fitness sharing, 1 a decision in part made on the availability of SBB source code.…”

“…The Symbiotic bid-based framework for GP as applied to classification -hereafter SBB -utilizes two populations linked through symbiosis [11,5]. Specifically, a program (symbiont) population assumes a bid-based GP representation [10] in which each individual consists of a tuple c, b .…”