Electronic applications enabling pervasive artificial intelligence (AI) have unique and conflicting challenges. These include extreme energy efficiency, accuracy and continuous learning. We are developing a new generation of AI hardware architecture to address these challenges.

We have built our architecture on the principle of learning automata. Mikhail Tsetlin proposed the theory of learning automata in the 1960s. It has long been a basis for non-linear control systems. It allows for formulation of machine learning using propositional logic and game theory.

We are working with our collaborators at the University of Agder, Norway, led by Professor Ole-Christoffer Granmo. Together, we are enabling powerful new AI applications through our hardware innovations.