Question B.1 Consider design of an error-back-propagation neural network to approximately classify a two dimensional (with real-valued elements) pattern into three categories: Class 1 for patterns inside the circle with center at (2,2) and a radius equal to 1 . Class 2 for patterns inside the circle with center at (4,5) and a radius equal to 1. Class 3 for patterns outside both circles. Explain in detail the design steps, i.e. (a) choosing a network architecture, (b) Creating training, validation and the test sets, (c) Selecting a performance criterion, deciding on the learning algorithm, testing and validation