Adaptive-learning neuron circuits using ferroelectric thin films

Abstract

Adaptive-learning neuron circuits are reviewed, in which a pulse frequency modulation (PFM) system is used and the interval of output pulses is changed through the learning process. Key devices of the circuits are MFSFETs (Metal Ferroelectric Semiconductor Field Effect Transistors). They are used for representing the synaptic weights of neurons and the polarity of the films is gradually changed by applying input pulses to the gates. In order to produce PFM signals, circuit using a UJT (unijunction transistor) is discussed. As a preliminary experiment for realization of MFSFETs, the electrical properties of metal ferroelectric metal (MFM) capacitor using sol-gel derived PZT films are discussed.