Title :
Neurotransistor:≫MINUSa neuron-like high-functionality transistor implementing intelligence on silicon
Author :
Shibata, Tadashi ; Ohmi, Tadahiro
Author_Institution :
Dept. of Electron. Eng., Tohoku Univ., Sendai, Japan
Abstract :
The main theme of this paper is not a so-called neural network but a new-architecture intelligent electronic circuit implemented using the neuron-like high-functionality transistor as a basic circuit element. The transistor is a multiple-input-gate thresholding device and is called neuron MOSFET (neuMOS or v MOS for short) due to its functional similarity to a brain cell neuron. The vMOS circuits are characterized by a high degree of parallelism in hardware computation, a large flexibility in hardware configuration and a dramatic reduction in the circuit complexity as compared to conventional integrated circuits. As a result, a number of new concept circuits have been developed. As examples, a real-time reconfigurable logic circuit called flexware and an associative memory conducting fully-parallel search for the most similar are presented. The enhancement in the functionality at a very elemental transistor level is critically important in building human-like intelligent systems on silicon
Keywords :
MOSFET; neural chips; Si; associative memory; flexware; fully-parallel search; integrated circuit; intelligent electronic circuit; most similar; multiple-input-gate thresholding device; neuron MOSFET; neurotransistor; real-time reconfigurable logic circuit; silicon; Biological neural networks; Brain cells; Concurrent computing; Electronic circuits; Flexible printed circuits; Hardware; Intelligent networks; MOSFET circuits; Neurons; Parallel processing;
Conference_Titel :
VLSI Signal Processing, VIII, 1995. IEEE Signal Processing Society [Workshop on]
Conference_Location :
Sakai
Print_ISBN :
0-7803-2612-1
DOI :
10.1109/VLSISP.1995.527474