Title :
A network architecture for maximum entropy estimation
Author :
Desilva, Christopher J S ; Choong, Poh Lian
Author_Institution :
Dept. of Electr. & Electron. Eng., Western Australia Univ., Nedlands, WA, Australia
fDate :
27 Jun-2 Jul 1994
Abstract :
This paper describes an artificial neural network (ANN) architecture for constructing maximum entropy models based on discrete distributions. Entropy is maximized by a constrained gradient ascent algorithm, which is shown to be capable of implementation by an ANN architecture. The use of this architecture as a method of inference is illustrated by applying it to a simple problem in probability theory
Keywords :
estimation theory; maximum entropy methods; neural net architecture; probability; artificial neural network architecture; constrained gradient ascent algorithm; discrete distributions; maximum entropy estimation; probability theory; Artificial intelligence; Artificial neural networks; Entropy; Image reconstruction; Inference algorithms; Information processing; Intelligent systems; Medical expert systems; Nonlinear equations; Probability distribution;
Conference_Titel :
Neural Networks, 1994. IEEE World Congress on Computational Intelligence., 1994 IEEE International Conference on
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-1901-X
DOI :
10.1109/ICNN.1994.374497
