Physical Review E (2000), 62:4010-4013
Frank C. Hoppensteadt and Eugene M. Izhikevich
Systems Science Center, Box 7606,
Arizona State University,
Tempe, AZ 85287-7606.
Abstract. We investigate here possible neuro-computational features of networks of laser oscillators. Our approach is similar to classical optical neurocomputing where artificial neurons are lasers and connection matrices are holographic media. However, we consider oscillatory neurons communicating via phases rather than amplitudes. Memorized patterns correspond to synchronized states where the neurons oscillate with equal frequencies and with prescribed phase relations. The mechanism of recognition is related to phase locking. We consider both solid-state and semiconductor lasers and discuss their advantages and drawbacks for this neurocomputing paradigm.
Full text in PDF file ,