An Introduction to Back Propagation Learning and its Application in Classification of Genome Data Sequence

Abstract

The gene classification problem is still active area of research because of the attributes of the genome data, high dimensionality and small sample size. Furthermore, the underlying data distribution is also unknown, so nonparametric methods must be used to solve such problems. Learning techniques are efficient in solving complex biological problems due to characteristics such as robustness, fault tolerances, adaptive learning and massively parallel analysis capabilities, and for a biological system it may be employed as tool for data-driven discovery. In this paper, some concepts related to cognition by examples are discussed. A classification technique is proposed in which DNA sequence is analyzed on the basis of sequence characteristics near breakpoint that occur in leukemia. The training dataset is built for supervised classifier and on the basis of that back propagation learning classifier is employed on hypothetical data. Our intension is to employ such techniques for further analysis and research in this domain. The future scope and investigation is also suggested.