Parallel Processing Through Hybrid Photonic Inference Machines

The goal of the proposed project is to develop a hybrid photonic inference machine to exploit the latent parallelism whic often underlines inferential or deductive processes. The resulting technology will not only serve to accelerate logical programs, but also other inferential processes involving search. In fact, a basic theme of artificial intelligence is that search is the foundation of intelligence. Hence the proposed project has wide-ranging implications for artificial intelligence and its applications to practical systems. Moreoever, the project is to such scope that a number of derivative applications are expected. For instance, a key aspect of the proposed work lies in parallel search through a society of cooperating modules. The resulting technology will therefore be useful for any distributed computational process, including parallel search through a large network of databases. ANTICIPATED BENEFITS: A high performance inference machine will have wide-ranging applications in any domain where high-level reasoning is paramount. Such applications occur in all fields of commerce, industry, and science, as demonstrated by previous artifical intelligence programs in these fields. Moreover, the aspect of the work on parallel search will be applicable to many disciplines and industries, from outline libraries to image processing, and from commercial customer lists to industrial data sets.