1. Using Hebb learning, construct a neural network that will recognize the first five letters of your

first name. If you dont have five letters in your first name than remaining letters should be

taken from the last name. Use a 5X5 grid and assume that the letters are in caps. Your input

layer will have twenty six neurons and output layer will have 5 neurons. You should have five

iterations of Hebb learning. example name hari

2. What is back-propagation? List the parameters that govern convergence in the back propagation algorithm.

3. What does resonance mean in the ART (Adaptive Resonance Theory)? What role does the

vigilance parameter play in an ART network? What is the difference between ART1 and

ART2?

4. Consider an ART1 neural net with four F1 units and three F2 units. An initial training of ART1 net yields bij = (0.67, 0.0, 0.2, 0.0, 0.0, 0.2, 0.0, 0.0, 0.2, 0.0, 0.67, 0.2) and tji = (1, 0, 0, 0, 0,0, 0, 1, 1, 1, 1, 1). Assuming vigilance factor to be 0.7, determine the weight matrix after the vector (0, 0, 0, 1) is presented for training.

5. Consider an ART2 network with two input units (n=2). Show that using =0.7 will force the input patterns (0.71, 0.69) and (0.69, 0.71) to different clusters. What role does the vigilance parameter play in this situation?