A Neural Network Method for Learning Statistical Knowledge from Multihypothesis Observation Data

摘要

This paper presents a neural network method for learning the statistical knowledge (i.e., the a posteriori probability functions) from observation sequences under the condition that the a priori probability density functions of the observation data are not provided by the observation environment. A multilayer feedforward network architecture and a reinforcement learning algorithm are designed for learning the posterior probability functions from the observation data of multihypothesis sources. It is shown that the neural-network based estimator can closely approximate the target functions by the simulation results conducted in the paper.