Extensible neural network software : applications in gene expression analysis

Abstract

Artificial Neural Networks have been increasingly utilized in the life sciences for analysis of large data sets. High-throughput technologies, such as gene expression microarrays, have challenged traditional statistical learning algorithms given their high dimensionality. This thesis describes GAINN, a neural network software package I created. GAINN was designed to be an extensible tool for both researches and students to use in neural network explorations. Several algorithms and features were implemented and tested on classification of various gene expression array data sets. The code design and user interface were implemented in such a manner that new algorithms and features would be trivial to incorporate into GAINN.