The analysis step of the Knowledge Discovery in Databases or data mining often relies on classifiers like neural network algorithms for doing parallel data searches. The CogniMem device is a direct implementation of the RBF (Radial Basis Function), RCE (Restricted Coulomb Energy) and KNN non- linear classifiers. Implementing a search engine on an optimized hardware based solution versus algorithmically on von-Neumann will provide orders of magnitude of cost/operation/watt/second enhancement. Each CogniMem delivers 131 gops of performance in a .5 watt component that can be scaled without loss of performance.
KNN searching for K nearest neighbor (or vectors stored) in 10 usec
Financial services & Scientific computing