Cognition Improved Neural Network Scheme for Detecting Rare Motifs in DNA Sequences

The success of the learning theory of neural networks is affected by many factors. First and foremost is the representation of data. If the inputs presented to the neural network are much irrelevant then the knowledge discovery is very difficult. Preprocessing the training data supports for the cognitive ability of the network and this will definitely improve the training time of neural network. Data preprocessing is essential especially for problems in the field of Bioinformatics where there is a large amount of data taken for analysis. Providing suitably preprocessed input data to the networks may serve as a critical factor in the cognitive ability and processing power of the networks. Hence, an attempt has been made in this direction to construct an artificial neural network with the support of numerically preprocessed input data sets and the results are provided. The performances of the different network architectures are also compared. The product obtained from data preprocessing is the final training set. In this paper, the significance of preprocessing of the training data is discussed in order to improve the cognitive ability of the network which will make better accuracy.