Theory of the fuzzy controller
Fuzzy Sets and Systems
IEEE Spectrum
A comparison of fuzzy forecasting and Markov modeling
Fuzzy Sets and Systems
Properties of certain fuzzy linear regression methods
Fuzzy Sets and Systems
A probabilistic and statistical view of fuzzy methods
Technometrics
Fuzzy logic: better science? or better engineering?
Technometrics
Application of Fuzzy Logic to Approximate Reasoning Using Linguistic Synthesis
IEEE Transactions on Computers
Fuzzy basis functions, universal approximation, and orthogonal least-squares learning
IEEE Transactions on Neural Networks
Hi-index | 0.00 |
The analysis of gene expression microarrays plays an important role in elucidating the functionality of genes, including the discovery of genetic interactions that regulate gene expression. Several methods for modeling such gene regulatory networks exist, including a variety of continuous and discrete models. Methods based on fuzzy logic provide an interesting alternative. However, the guidelines for modeling gene expression with fuzzy logic are fairly open, and the need arises to investigate how adjustments in the modeling scheme will affect the results. In this work, we modify an existing fuzzy logic algorithm to involve an arbitrary number of classification states, and investigate the limiting behavior as the number of states tends to infinity. We also propose a probabilistic model as an alternative to the fuzzy logic model. We investigate the behavior of both models using yeast cell-cycle data and the simulated data of Werhli et al. [A.V. Werhli, M. Grzegorczyk, D. Husmeier, Comparative evaluation of reverse engineering gene regulatory networks with relevance networks, graphical Gaussian models, and Bayesian networks, Bioinformatics 22 (2006) 2523-2531]. We found that altering the number of classification states in both the fuzzy logic and probability models can influence which networks are predicted by both models. As the number of states tends to infinity, the predictions made by both models converge to those of a regression model. Models with a small to moderate number of classification states produced better results from a biological standpoint, compared to models with higher numbers of states. In simulated data, models with differing numbers of classification states produced similar overall results. Thus, increasing the complexity of the models has no apparent benefit, and models with smaller numbers of classification states are therefore preferred based on their ease of linguistic interpretation. The software used in this paper is freely available for non-commercial use at http://louisville.edu/~g0broc01.