tinyML EMEA Monday Keynote
ICL Professor of Computer Engineering, The University of Manchester
A Novel Mechanism for Edge ML
Sparse Binary Coincidence (SBC) memories store coincidences between features detected in class examples in a training set, and infer the class of a previously unseen test example by identifying the class with which it shares the highest number of feature coincidences. This mechanism supports single pass and continuous learning, robustness to imperfect inputs, and has no need for floating-point numbers or differentiation. As such it is ideally suited to implementation on small platforms with limited resources, such as those used for AI at the edge and those used for neuromorphic computation.