Summary: | 碩士 === 國立清華大學 === 電子工程研究所 === 86 === Over the last several decades, neuromophic scientists have long
sought to realize the way human brain works in a computer.
Various on-chip analog storage devices were designed to achieve
the biological functions of synapses or neurons. In this work we
propose a novel device structure for an electrically trainable
single-transistor silicon neuron, based on the Flash EEPROM
technology. This device contains multiple control gates and the
corresponding floating gates to specify the act of an electronic
neuron.Signals biased on the control gates are just like the
input excitations of synapses, and the charges on the
corresponding floating gates determine the weighting factors of
the synapses. We propose the basic structure and operations of
our device, including the analogy to a biological neuron and the
methods of updates with simulation results. Conclusions were
made with a simplified equivalent circuit model, and several
structural suggestions for future development are also
discussed.
|