Page 177 - Introduction to Information Optics

162 2. Signal Processing with Optics

( a ) ( c }

following subsets:
(a) I and II.
(b) 1 and V.
(C) VI and X.
(d) IX and XI.
(e) III and IX.
2.47 A set of the training patterns shown in Fig. 2.75 are to be stored in a
neural network.
(a) Use a Hopfield model to construct a one-level neuron network.
(b) With reference to part (a), construct the memory matrix.
(c) Determine the interconnections by logical operation.
(d) Construct a one-level IPA neural net.
(e) Compare the IPA model with the Hopfield model.
2.48 Refer to the set of training images shown in Fig. 2.76. Synthesize a
synthetic discriminant function filter (SDF).
2.49 Consider the target set of Fig. 2.76. Synthesize a simulated annealing
bipolar composite filter (SABCF).

Fig. 2.76.
2.50 Use Fig. 2.76 as the target set and Fig. 2.77 as the antitarget set, and
synthesize a SABCF. In view of the results obtained in Exercises 2.48 and
2.49, comment on the SDF and SABCF results.

Fig. 2.77.