Publications
Four peer-reviewed papers on predictive error compensation.
Four papers, 2020 to 2024 · Updated Sep 24, 2026
What do the papers describe?
A primary network predicts a time series from inputs at several time scales. A second network, trained separately, predicts the primary network’s error and corrects it.
Read the papers
Papers
High-Performance Time Series Prediction With Predictive Error Compensated Wavelet Neural Networks
2020IEEE Access
Stock Price Prediction Using Predictive Error Compensation Wavelet Neural Networks
May 6, 2021Computers, Materials & Continua
Predictive Error Compensating Wavelet Neural Network Model for Multivariable Time Series Prediction
2021TEM Journal
Improvement in Prediction Performance Using Predictive Error Compensated Neural Networks
Oct 1, 2024Springer
Partnership
The papers come from an international research partnership through ARRTECH Türkiye.
Names
The method first appeared as PEC-WNN and later as PECNET.
Scope
The papers test prediction on stock prices, market indices, wind speed and the Lorenz attractor. None uses security telemetry.
Citations
Ustundag and Kulaglic, IEEE Access 8, 210532–210541, 2020
Kulaglic and Ustundag, Computers, Materials & Continua 68(3), 3577–3593, 2021
Kulaglic and Ustundag, TEM Journal 10(4), 1955–1963, 2021
Kulaglic, Advanced Technologies, Systems, and Applications IX, LNNS 1143, 186–194, Springer 2024
Related
Graph neural networks
Method
Unsupervised behavioral modeling
Method
Evaluations and benchmarks
Method
Schedule a meeting
