The Trainable Inference Network (TIN) is an adaptive network that can learn the functionality of any finite-state machine. TIN is composed of two modified adaptive resonance circuits (ARCs) that learn transition and output tables and an auxiliary assembly of control nodes that facilitates state transitions. The first ARC learns to recognize current-state, input, next-state patterns appearing on separate slabs; the other learns current-state, input, output patterns. Features of TIN's macrocircuit and dynamics are described, focusing on the first, state-transition ARC. TIN is then taught the states, input, and transitions of two simple finite-state machines. The results are summarized, and future research directions are indicated.
|Original language||English (US)|
|State||Published - 1987|
ASJC Scopus subject areas