论文信息 - Silicon implementation of an auto-adaptive network for real-time separation of independent signals

Silicon implementation of an auto-adaptive network for real-time separation of independent signals

A nonlinear autoadaptive filter for the separation of independent signal sources is described. The filter takes the form of a recurrent network which uses N simple processing units interconnected by N(N-1) inhibitory synapses. Each synapse determines its weight from a local Hebbian-like learning rule. Criteria are specified which the learning rule must satisfy, and examples from testing with digital simulations are given. The network was implemented using analog VLSI technology so as to achieve real-time, low-power, scalable solutions. Test results from a 2*/sub /chip are presented.<<ETX>>

[1] Eric A. Vittoz,et al. CMOS Integration of Herault-Jutten Cells for Separation of Sources , 1989, Analog VLSI Implementation of Neural Systems.

[2] Andreas G. Andreou,et al. Current-Mode Subthreshold MOS Circuits for Analog , 1991 .

[3] D. Brillinger. Time Series: Data Analysis and Theory , 1981 .

[4] Andreas G. Andreou,et al. Current-mode subthreshold MOS circuits for analog VLSI neural systems , 1991, IEEE Trans. Neural Networks.

[5] C. Jutten. IN. C. A. : An independenct components analyser based on an adaptive neuromimetic network , 1988 .

[6] J. Demongeot,et al. Artificial intelligence and cognitive sciences , 1988 .

[7] Michel Steyaert,et al. Analog VLSI implementation of Neural Networks , 1992 .

[8] Carver A. Mead. Analog VLSI and neural systems (invited presentation) , 1989 .

[9] Carver Mead,et al. Analog VLSI and neural systems , 1989 .

[10] Tobi Delbriick. "Bump" Circuits for Computing Similarity and Dissimilarity of Analog Voltages , 1991 .