Adaptive Learning with Covariate Shift-Detection for Non-Stationary Environments

Haider Raza, G Prasad, Yuhua Li

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Citations (Scopus)

Abstract

Learning with data-set shift is a major challenge in non-stationary environments wherein the input data distribution may shift over time. Detecting the data-set shift point in the time-series data, where the distribution of time-series shifts its properties, is of utmost interest. Data-set shift exists in a broad range of real-world systems. In such systems, there is a need for continuous monitoring of the process behavior and tracking the state of the shift so as to decide about initiating adaptation in a timely manner. This paper presents an adaptive learning algorithm with data-set shift-detection using an exponential weighted moving average (EWMA) model based test in a non-stationary environment. The proposed method initiates the adaptation by re-configuring the knowledge-base of the classifier. This algorithm is suitable for real-time learning in non-stationary environments. Its performance is evaluated through experiments using synthetic data-sets. Results show that it reacts well to different co-variate shifts.
LanguageEnglish
Title of host publicationUnknown Host Publication
Number of pages8
DOIs
Publication statusPublished - 8 Sep 2014
Event2014 14th UK Workshop on Computational Intelligence (UKCI) - Bradford, UK
Duration: 8 Sep 2014 → …

Workshop

Workshop2014 14th UK Workshop on Computational Intelligence (UKCI)
Period8/09/14 → …

Fingerprint

Time series
Adaptive algorithms
Learning algorithms
Classifiers
Monitoring
Experiments

Cite this

@inproceedings{5c41077347c94b05a1f97b3e04412069,
title = "Adaptive Learning with Covariate Shift-Detection for Non-Stationary Environments",
abstract = "Learning with data-set shift is a major challenge in non-stationary environments wherein the input data distribution may shift over time. Detecting the data-set shift point in the time-series data, where the distribution of time-series shifts its properties, is of utmost interest. Data-set shift exists in a broad range of real-world systems. In such systems, there is a need for continuous monitoring of the process behavior and tracking the state of the shift so as to decide about initiating adaptation in a timely manner. This paper presents an adaptive learning algorithm with data-set shift-detection using an exponential weighted moving average (EWMA) model based test in a non-stationary environment. The proposed method initiates the adaptation by re-configuring the knowledge-base of the classifier. This algorithm is suitable for real-time learning in non-stationary environments. Its performance is evaluated through experiments using synthetic data-sets. Results show that it reacts well to different co-variate shifts.",
author = "Haider Raza and G Prasad and Yuhua Li",
year = "2014",
month = "9",
day = "8",
doi = "10.1109/UKCI.2014.6930161",
language = "English",
booktitle = "Unknown Host Publication",

}

Raza, H, Prasad, G & Li, Y 2014, Adaptive Learning with Covariate Shift-Detection for Non-Stationary Environments. in Unknown Host Publication. 2014 14th UK Workshop on Computational Intelligence (UKCI), 8/09/14. https://doi.org/10.1109/UKCI.2014.6930161

Adaptive Learning with Covariate Shift-Detection for Non-Stationary Environments. / Raza, Haider; Prasad, G; Li, Yuhua.

Unknown Host Publication. 2014.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Adaptive Learning with Covariate Shift-Detection for Non-Stationary Environments

AU - Raza, Haider

AU - Prasad, G

AU - Li, Yuhua

PY - 2014/9/8

Y1 - 2014/9/8

N2 - Learning with data-set shift is a major challenge in non-stationary environments wherein the input data distribution may shift over time. Detecting the data-set shift point in the time-series data, where the distribution of time-series shifts its properties, is of utmost interest. Data-set shift exists in a broad range of real-world systems. In such systems, there is a need for continuous monitoring of the process behavior and tracking the state of the shift so as to decide about initiating adaptation in a timely manner. This paper presents an adaptive learning algorithm with data-set shift-detection using an exponential weighted moving average (EWMA) model based test in a non-stationary environment. The proposed method initiates the adaptation by re-configuring the knowledge-base of the classifier. This algorithm is suitable for real-time learning in non-stationary environments. Its performance is evaluated through experiments using synthetic data-sets. Results show that it reacts well to different co-variate shifts.

AB - Learning with data-set shift is a major challenge in non-stationary environments wherein the input data distribution may shift over time. Detecting the data-set shift point in the time-series data, where the distribution of time-series shifts its properties, is of utmost interest. Data-set shift exists in a broad range of real-world systems. In such systems, there is a need for continuous monitoring of the process behavior and tracking the state of the shift so as to decide about initiating adaptation in a timely manner. This paper presents an adaptive learning algorithm with data-set shift-detection using an exponential weighted moving average (EWMA) model based test in a non-stationary environment. The proposed method initiates the adaptation by re-configuring the knowledge-base of the classifier. This algorithm is suitable for real-time learning in non-stationary environments. Its performance is evaluated through experiments using synthetic data-sets. Results show that it reacts well to different co-variate shifts.

U2 - 10.1109/UKCI.2014.6930161

DO - 10.1109/UKCI.2014.6930161

M3 - Conference contribution

BT - Unknown Host Publication

ER -