TY - JOUR
T1 - Integrating adaptive moving window and just-in-time learning paradigms for soft-sensor design
AU - Urhan, Aysun
AU - Alakent, Burak
PY - 2020
Y1 - 2020
N2 - Most applications of soft sensors in process industries require learning from a stream of data, which may exhibit nonstationary dynamics, or concept drift. In this study, we develop a relevance vector machine (RVM) based novel adaptive learning algorithm called MWAdp-JITL, to meet the demands of continuous processes. The resulting algorithm combines active and passive learning: A moving window (MW) algorithm, which adapts the window size against virtual/real concept drifts, is coupled with a just-in-time learning (JITL) model, constructed using an appropriate region of historical data, and the ensemble weights of the MW and JITL models are adjusted for each query point. Tests on four real industrial datasets and a synthetic data, comprising various concept drift scenarios, show that MWAdp-JITL yields superior prediction accuracy and is generally more robust to changes in algorithm parameters compared to conventional adaptive learning methods and state-of-the-art algorithms from the literature. MWAdp-JITL complies with time limits of online prediction, and is applicable for high dimensional processes under various types of concept drifts. It is seen that MWAdp-JITL can successfully achieve a good balance in bias-variance tradeoff, justifying the use of only two exquisitely selected learners in ensemble learning.
AB - Most applications of soft sensors in process industries require learning from a stream of data, which may exhibit nonstationary dynamics, or concept drift. In this study, we develop a relevance vector machine (RVM) based novel adaptive learning algorithm called MWAdp-JITL, to meet the demands of continuous processes. The resulting algorithm combines active and passive learning: A moving window (MW) algorithm, which adapts the window size against virtual/real concept drifts, is coupled with a just-in-time learning (JITL) model, constructed using an appropriate region of historical data, and the ensemble weights of the MW and JITL models are adjusted for each query point. Tests on four real industrial datasets and a synthetic data, comprising various concept drift scenarios, show that MWAdp-JITL yields superior prediction accuracy and is generally more robust to changes in algorithm parameters compared to conventional adaptive learning methods and state-of-the-art algorithms from the literature. MWAdp-JITL complies with time limits of online prediction, and is applicable for high dimensional processes under various types of concept drifts. It is seen that MWAdp-JITL can successfully achieve a good balance in bias-variance tradeoff, justifying the use of only two exquisitely selected learners in ensemble learning.
KW - Adaptive learning
KW - Concept drift
KW - Continuous process
KW - Ensemble learning
UR - http://www.scopus.com/inward/record.url?scp=85079267624&partnerID=8YFLogxK
U2 - 10.1016/j.neucom.2020.01.083
DO - 10.1016/j.neucom.2020.01.083
M3 - Article
AN - SCOPUS:85079267624
SN - 0925-2312
VL - 392
SP - 23
EP - 37
JO - Neurocomputing
JF - Neurocomputing
ER -