CMOS temporal associative memory

Abstract

In this paper we present a mixed digital analog approach for VLSI implementation of an associative memory model using temporal relations. The proposed model is based on the biological model of the cortex. There are two motivations for this research. First, the analog and the parallel nature of the neural network approach may provide an efficient technique to achieve the high speed requirement for real time coding systems with less hardware than both digital techniques and adaptive neural techniques. Second, the model proposed based on the biological neural network may be useful as a model of the information processing in human brain. The proposed model overcomes the drawbacks of the linear associative memory. The proposed circuit realizing such a theory is faster, smaller in area, and more efficient than the current systems.