Systems And Methods For Providing Programmable Analog Classifiers
Halser, Paul E.
MetadataShow full item record
The present invention describes systems and methods to provide programmable analog classifiers. An exemplary embodiment of the present invention provides an analog classifier circuit comprising a bump circuit enabled to store a template vector, wherein the template vector can model a probability distribution with exponential behavior. Furthermore, the bump circuit is enabled to generate an output corresponding to a comparison between an input vector received by the bump circuit and the template vector stored by the bump circuit. Additionally, the analog classifier circuit includes a variable gain amplifier in communication with the bump circuit, and the variable gain amplifier can be adjusted to modify the variance of the template vector.
- Georgia Tech Patents