A neural network model for adaptive, non-uniform A/D conversion

Author

Hulle, Marc M Van

Author_Institution

Dept. of Brain & Cognitive Sci., MIT, Cambridge, MA, USA

fYear

1993

fDate

6-9 Sep 1993

Firstpage

555

Lastpage

561

Abstract

An adaptive feedforward network is presented for performing non-uniform, flash-type analog-to-digital (A/D) conversion. The unsupervised competitive learning rule used, called boundary adaptation rule (BAR), maximizes entropy and provides an efficient nonuniform quantization of the analog signal range. The network is easily implementable in VLSI circuitry and meets the requirements of smart sensors. It is shown that the network is able to adapt itself to rapidly changing input signals, such as speech signals

Keywords

analogue-digital conversion; feedforward neural nets; maximum entropy methods; signal processing; unsupervised learning; adaptive feedforward network; boundary adaptation rule; maximum entropy; neural network model; nonuniform A/D conversion; nonuniform quantization; smart sensors; speech signals; unsupervised competitive learning; Adaptive systems; Biological neural networks; Circuits; Entropy; Intelligent sensors; Neural networks; Neurons; Quantization; Sensor phenomena and characterization; Very large scale integration;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks for Processing [1993] III. Proceedings of the 1993 IEEE-SP Workshop

Conference_Location

Linthicum Heights, MD

Print_ISBN

0-7803-0928-6

Type

conf

DOI

10.1109/NNSP.1993.471832

Filename

471832