We put forward a novel chaos neuron model and investigate its dynamic properties. The artificial neuron model is defined by such periodic input-output mapping as a sinusoidal function. It is obviously shown that the present neural network has the ability of embedded pattern retrieval in the neural network beyond the conventional one with a monotonous mapping such as the signum function. This advantage is considered to be as a result of the nonmonotonous property of the periodic mapping which involves the chaotic behaviour of the neurons. It is also found that the present chaos neuron model has a remarkably larger memory capacity than the conventional association system with the monotonous dynamics.
Index Terms:
neural nets; chaos; content-addressable storage; transfer functions; chaos neural network; periodic activation function; artificial neuron model; periodic input-output mapping; sinusoidal function; embedded pattern retrieval; monotonous mapping; signum function; periodic mapping; memory capacity; association system; associative memory
Citation:
M. Nakagawa, "A study of chaos neural network with a periodic activation function," annes, pp.45, 2nd New Zealand Two-Stream International Conference on Artificial Neural Networks and Expert Systems (ANNES '95), 1995
PURCHASE ARTICLE: $19
IEEE Xplore Subscribers
Digg
Furl
Spurl
Blink
Simpy
Google
Del.icio.us
Y!MyWeb