Neural Networks For Dummies And In Everyday Life

Anne Menendez, Silicon Recognition, Petaluma

Stevenson Hall 1300
12:00 PM - 12:50 PM

Neural networks are known for their ability to solve fuzzy and ill-defined problems that are too complex for conventional technologies. Understanding neural networks is also known to require experience in mathematical modeling and computer programming, not to mention the need for powerful computer systems. They have been used in R&D and governmental laboratories for years but are slowly emerging in applications such asfinancial risk assessment, robotics, machine vision, predictive maintenance and more. Anew neural network silicon chip called ZISC (Zero Instruction Set Computing) has the capability to turn this once privileged technology into an applied, easy-to-use and affordable technology. The chip implements the known RBF (Radial Basis Function) and KNN (K-Nearest Neighbor) neuronal models, but its key feature is a parallel architecture which delivers ultra high-speed performance and unlimited expandability. In consequence, applications of the ZISC can be seen in workstations (i.e. brainputer) for massive real-timedata mining where one pattern has to be matched among millions in a few milliseconds, as well as in appliances and sensors where intelligence (i.e. pattern recognition or non-recognition) can be distributed locally for direct decision making or for the selective transmission of the information of interest.