NOISE PERFORMANCE OF LINEAR ASSOCIATIVE MEMORIES

The performance of two commonly used linear models of associative memo ries, generalized inverse (GI) and correlation matrix memory (CMM) is studied analytically in the presence of a new type of noise (training noise due to noisy training patterns). Theoretical expressions are det ermined for the SNR (signal-to-noise ratio) gain of the GI and CMM mem ories in the auto-associative and hetero-associative modes of operatio n. It is found that the GI method performance degrades significantly i n the presence of training noise while the CMM method is relatively un affected by it. The theoretical expressions are plotted and compared w ith the results obtained from Monte Carlo simulations and the two are found to be in excellent agreement.