Volatility from Threshold Noise in Randomly Connected Synchronous Asymmetric Neural Networks

P.C. McGuire, J. Rafelski, C. Pershing, H. Bohr

We study the diversity of complex spatio-temporal patterns of random synchronous asymmetric neural networks (RSANNs). Specifically, we investigate the impact of noisy thresholds on network performance and find that there is an interesting region of noise parameters where RSANNs display specific features of behavior desired for rapidly responding processing systems: accessibility to a large set of distinct, complex patterns. Such accessibility may one day lead to `creative' RSANNs.