Bipolar pattern association using a two-layer feedforward neural network

The authors present a design technique that constructs a two-layer feedforward network for the realization of an arbitrary set of bipolar associations (p_i, q_i), i-=1, ..., k. The underlying idea is to use a layer of the hard-logic neurons to identify each p_i in the winner-take-all fashion. Then, the second layer of the so-called sign neurons picks up the corresponding pattern q _i. An important feature of the net is that it can be used as an error-correcting associative memory if the thresholds of the hard-logic-neurons in the first layer are properly adjusted