A Neuroph Approach for Classifying Ecoli Data

Classification methods are used in handling classification problems, which usually exist when entities or objects needs to be assigned predefined groups or classes, perhaps based on attributes, parameters and values. This research paper provides a simplified description of neuroph classification using Ecoli data. The data were structured into training and testing datasets. The neuroph neural network architecture caters for 34 neurons: twelve input neurons (12), seventeen (17) hidden neurons and five (5) output neurons. The training accommodated approximately 185 iterations with a cumulative error of 6.8544 and an average error of 0.0367. The Total Mean Square Error (TMSE) obtained from testing the trained data gave an approximate value of 2.7372. This minima error showed the optimality in training and classification using Ecoli data. General Terms Neural network, artificial neural network, artificial intelligence, neurons