Learning Under Concept Drift for Regression—A Systematic Literature Review

Abstract

Context: The amount and diversity of data have increased drastically in recent years. However, in certain situations, the data to which a trained Machine Learning model is significantly different from testing data, a problem known as Concept Drift (CD). Because CD can be a serious issue, there has been a wealth of research on how to detect and work around it. However, most of the literature focuses on classification tasks. Objective: Making a Systematic Literature Review (SLR) for CD in the context of regression. Research questions: How to detect CD and how to build CD techniques for regression problems using machine learning? Method: We ran an automatic search process on reference databases, selecting papers from 2010 to August 2020, following the methodological process proposed by (Kitchenhame and Charters) (2007). Results:We selected 41 papers. Drift Detection Methods based on ensembles and neural networks with highlight OS-ELM were the most frequent in the selected papers with superior performance. However, only two papers confirm such superiority statistically. Furthermore, identify CD problems as the batch size, drift points, and where drift happens. Conclusions: SLR focuses on highlighting the existing literature on CD applied to regression.

Concept Drift Adaptation Methods under the Deep Learning Framework: A Literature Review

“…First, the data stream input (single input or batch input) is generally trained and learned by the deep learning model (single model or ensemble model) to obtain the basic prediction results. Next, if concept drift occurs during this process, a concept drift adaptation method will be triggered to update the deep learning model to accommodate concept drift and maintain the model [16,19]. The concept drift adaptation method can be divided into two parts: concept drift detection and model update.…”

“…The changes in concept may manifest in different forms over time. At present, the most popular types of concept drift can be divided into abrupt drift, incremental drift, gradual drift, and recurring drift [14,16,17].…”

“…Ghani et al [15] mainly introduced conceptual drift detection methods for untagged data streams. Lima et al [16] summarized and classified concept drift adaptation methods based on different machine learning models under the regression task. Bayram et al [17] classified and reviewed the existing detection algorithms of concept drift adaptation methods.…”

A feature selection method based on Shapley values robust for concept shift in regression

“…Much of the related research in this field focuses on shift detection (whether shift occurs or not), on understanding why it occurs (when, how and where) and on shift adaptation (reacting to change). It is also often treated from the perspective of classification problems, while the field of regression has not been explored in great depth [25]. Most strategies designed to react to the presence of the shift are based on retraining the models with more current data or with data similar to those occurring in the current context, although other strategies are possible [23].…”

A Novel Edge Architecture and Solution for Detecting Concept Drift in Smart Environments

“…[17]. In addition, most of the literature on concept drift focuses on classification tasks, whereas regression problems still require exploration [18]. Moreover, use cases resembling smart city applications with large datasets rarely incorporate concept drift detection and adaptation, which requires more exploration and experimentation [13].…”