Feedforward, high density, programmable read only neural network based memory systemNeural network-inspired, nonvolatile, programmable associative memory using thin-film technology is demonstrated. The details of the architecture, which uses programmable resistive connection matrices in synaptic arrays and current summing and thresholding amplifiers as neurons, are described. Several synapse configurations for a high-density array of a binary connection matrix are also described. Test circuits are evaluated for operational feasibility and to demonstrate the speed of the read operation. The results are discussed to highlight the potential for a read data rate exceeding 10 megabits/sec.
Document ID
19890027088
Acquisition Source
Legacy CDMS
Document Type
Conference Paper
Authors
Daud, Taher (Jet Propulsion Lab., California Inst. of Tech. Pasadena, CA, United States)
Moopenn, Alex (Jet Propulsion Lab., California Inst. of Tech. Pasadena, CA, United States)
Lamb, James (Jet Propulsion Lab., California Inst. of Tech. Pasadena, CA, United States)
Thakoor, Anil (Jet Propulsion Lab., California Inst. of Tech. Pasadena, CA, United States)
Khanna, Satish (California Institute of Technology Jet Propulsion Laboratory, Pasadena, United States)