File(s) under permanent embargo
Neural Network Training for Uncertainty Quantification over Time-Range
journal contribution
posted on 2022-10-25, 22:34 authored by Hussain Mohammed Dipu Kabir, Abbas KhosraviAbbas Khosravi, Saeid Nahavandi, D SrinivasanTraditional uncertainty quantification (UQ) algorithms are mostly developed for a fixed time (term), such as hourly or daily predictions. Although a few UQ techniques can compute UQ over time-range, their quantified uncertainty is usually ever-increasing and non-smooth. However, uncertainty can be lower at a certain time in the future. Therefore, this paper presents a neural network (NN) training procedure for both short-term and long-term uncertainty quantification to investigate the level of uncertainty over different times. The training procedure is similar to the conventional lower-upper bound estimation (LUBE) method. The proposed input combination consists of traditional input components and the term of the prediction. The proposed output is the UQ for a sample at that term. Estimation of sub-sample value, initial training with rough targets, and quality balancing over different term-range results in faster training and uniformity. According to the outputs of trained NNs, the uncertainty increases from very short-term to the short term but the uncertainty may decrease in the midterm or the long term. Moreover, the uncertainty may have a periodic portion over time in the long term. We also provide explanations of such periodicity in uncertainty over time curves.
History
Journal
IEEE Transactions on Emerging Topics in Computational IntelligenceVolume
5Pagination
768 - 779Publisher DOI
eISSN
2471-285XUsage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC