High-order optical associative memory based on neural nets

Kathleen E. Hamilton, Jonathan Schrock, Neena Imam +1 more

Gerasimos G. Rigatos, Spyros G. Tzafestaş

Chu Kiong Loo, Mitja Peruš, Horst Bischof

A quantum associative memory, much more natural than those of “quantum computers”, is presented. Neural-net-like processing with real-valued variables is transformed into processing with quantum waves. Successful computer simulations of image storage and retrieval are reported. Our Hopfield-like algorithm allows quantum implementation with holographic procedure using present-day quantum-optics techniques. This brings many advantages over classical Hopfield neural nets and quantum computers with logic gates.

Chu Kiong Loo, Mitja Peruš, Horst Bischof

It is computationally demonstrated how quantum associative networks, implemented using quantum holography, could be harnessed for object recognition. These simulated quantum nets alone execute efficient image recognition, i.e., reconstruction of an image selected from associative memory (hologram). However, optically implementable neural-net preprocessing of object-images is needed for appearance-based viewpoint-invariant recognition of objects. We present computer simulation results of two methods: Moore-Penrose orthogonalization and encoding of object-images with Gabor wavelets. A computer-supported quantum Gabor-wavelet holography is proposed. © 2005 Optical Society of America

Gregory R. Steinbrecher, Jonathan P. Olson, Dirk Englund +1 more

Dana Z. Anderson

A high capacity associative memory would find applications in computing and perhaps lend some insight into the understanding of collective neural networks and human memory. In this work an entirely optical associative memory device is proposed. The device operates on the principle that an optical resonator employing a holographic grating can have user prescribed eigenmodes. Information which is to be recalled is contained in the hologram. Each information entity (e.g., an image of a cat) defines an eigenmode of the resonator. The stored information is accessed by injecting partial information (e.g., an image of the cat’s ear) into the resonator. The appropriate eigenmode is selected through a competitive process in a gain medium placed inside the resonator. With a net gain greater than one, the gain amplifies the field belonging to the eigenmode which most resembles the injected field; the other eigenmodes are suppressed via the competition for the gain. The principles underlying the proposed device are outlined and the issues which must be addressed are delineated.