Credit Assignment Among Neurons in Co-evolving Populations

Abstract

Abstract. Different credit assignment strategies are investigated in a two level co-evolutionary model which involves a population of Gaussian neurons and a population of radial basis function networks consisting of neurons from the neuron population. Each individual in neuron population can contribute to one or more networks in network population, so there is a two-fold difficulty in evaluating the effectiveness (or fitness) of a neuron. Firstly, since each neuron only represents a partial solution to the problem, it needs to be assigned some credit for the complete problem solving activity. Secondly, these credits need to be accumulated from different networks the neuron participates in. This model, along with various credit assignment strategies, is tested on a classification (Heart disease diagnosis problem from UCI machine learning repository) and a regression problem (Mackey-Glass time series prediction problem).

Many-Objective Cooperative Co-evolutionary Feature Selection: A Lexicographic Approach

“…Since the wrapper method is based on an CCEA, potential solutions of the problem will be evolved within populations of different species. In this case, a hybrid single-level and two-level approach is proposed [16]. One species will be used to evolve the parameters of the classifier while the features of the input dataset will also be split into several species.…”

Evolutionary Multi-objective Optimization in Uncertain Environments

“…In fact, there may be little correlation between diversity in the two feature spaces. Examples include niche sharing [84,123,243], crowding [61], clustering [50,300], lateral interference [155], Pareto potential regions [113] and k-th nearest neighbor [2,298]. Distance-based assessments is based on the relative distance between individuals in the feature space.…”

A coevolutionary algorithm for rules discovery in data mining

“…Coevolution algorithms can be implemented at different levels, i.e. single-level or two-level coevolutionary, depending on the type of module to be evolved simultaneously (Khare et al 2004). In single-level coevolution De Jong 1994, 2000), each evolving subpopulation represents a subcomponent of the problem to be solved and in two-level coevolution, system and modules are simultaneously optimized in separate subpopulations (Moriarty 1997, Khare et al 2004.…”