Thumbnail
Access Restriction
Open

Author Noone, Greg
Source CiteSeerX
Content type Text
File Format PDF
Language English
Subject Domain (in DDC) Computer science, information & general works ♦ Data processing & computer science
Subject Keyword Individual Pulse Train ♦ Stock-market Type Problem ♦ Good Tracking ♦ Time Series Network Predictor ♦ Post-deinterleaving Radar Pulse Train Problem ♦ Revealing Example ♦ Discontinuous Mode Change ♦ Radar Pulse Train Parameter Estimation ♦ Spurious Pulse ♦ Simple Recurrent Backpropagation Neural Network ♦ Introduction Multi-layered Feedforward Neural Network ♦ Radar Problem ♦ Considerable Success ♦ Neural Network ♦ Network Update ♦ Multiple Level Stagger ♦ Time Series Predictor ♦ Novel Hueristic Adaptive Error ♦ Chaotic Analytic Function
Description The post-deinterleaving radar pulse train problem requires estimation of the parameters and tracking of the individual pulse trains. A simple recurrent backpropagation neural network is used based on a simple state space time series formulation of the radar problem. The network incorporates a novel hueristic adaptive error threshold that allows simultaneously good tracking and parameter estimating abilities. Two simple but revealing examples are presented to show how the network is robust to missing and spurious pulses, as well as multiple level staggers with discontinuous mode changes. Introduction Multi-layered feedforward neural networks have met with considerable success when used as time series predictors for well known chaotic analytic functions as well as some "stock-market" type problems [1, 2]. Time series network predictors differ from function interpolating networks in that they involve the processing of patterns that evolve over time. If the time between network updates re...
Educational Role Student ♦ Teacher
Age Range above 22 year
Educational Use Research
Education Level UG and PG ♦ Career/Technical Study
Learning Resource Type Article
Publisher Date 1995-01-01
Publisher Institution In Proc. IEEE ANNES'95. IEEE