Characterizing concept drift

By Geoffrey I. Webb, Roy Hyde, Hong Cao, Hai-Long Nguyen, Franccois Petitjean. 2015

511 citations recorded by the source corpus.

Most machine learning models are static, but the world is dynamic, and increasing online deployment of learned models gives increasing urgency to the development of efficient and effective mechanisms to address learning in the context of non-stationary distributions, or as it is commonly called concept drift. However, the key issue of characterizing the different types of drift that can occur has not previously been subjected to rigorous definition and analysis. In particular, while some qualitative drift categorizations have been proposed, few have been formally defined, and the quantitative descriptions required for precise and objective understanding of learner performance have not existed. We present the first comprehensive framework for quantitative analysis of drift. This supports the development of the first comprehensive set of formal definitions of types of concept drift. The formal definitions clarify ambiguities and identify gaps in previous definitions, giving rise to a new comprehensive taxonomy of concept drift types and a solid foundation for research into mechanisms to detect and address concept drift.

Subfields: Artificial Intelligence, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Management Science and Operations Research

Citations in the corpus

Incremental Learning in Online Scenario (2020, background): “Concept [29] in classification problems is defined as the joint distribution P (X,Y ) where X is the input data and Y represents target variable.”
Performative Prediction (2020, background): “However, studying the problem at this level of generality has led to a number of difficulties in creating a unified language and objective (Gama et al., 2014; Webb et al., 2016), an issue we circumvent by assuming that the population distribution is determined by the deployed classifier.”
From Concept Drift to Model Degradation: An Overview on Performance-Aware Drift Detectors (2022, background): “This type of drift is also called full-concept drift [48], Figure 6(d).”
A Survey on Ensemble Learning for Data Stream Classification (2017, background): “stream learning and concept drift adaptation [Tsymbal 2004; Žliobaitė 2010; Hoens et al. 2012; Gama et al. 2014; Lemaire et al. 2015; Ditzler et al. 2015; Barddal et al. 2016; Webb et al. 2016].”
Quantum machine learning (2016, unknown): “Concept drift is an essential problem in machine learning: it refers to shifts in the distribution that is being sampled and learned [ 27].”
Analyzing concept drift and shift from sample data (2018, methodology): “Webb et al. (2016) propose a number of quantitative measures for drift that provide gross summaries of the drift between two time points.”
Streaming Graph Neural Networks via Continual Learning (2020, background): “The dynamics give rise to some new patterns [28], referring to the neighborhood information not captured by the GNN, and at the same time, many existing patterns that have been learned by the GNN in the previous network still maintain.”