A Bayesian approach to abrupt concept drift

Abstract This paper proposes a model for estimating probabilities in the presence of abrupt concept drift. This proposal is based on a dynamic Bayesian network. As the exact estimation of the parameters is unfeasible we propose an approximate procedure based on discretizing both the possible probability values and the parameter representing the probability of change. The result is a method which is quite efficient in time and space (with a complexity directly related to the number of points used in the discretization) and providing very accurate predictions as well. These benefits are checked with a detailed comparison with other standard procedures based on variable size windows or forgetting rates.

• Publication year

  2019

• Venue

  Knowledge-Based Systems

• Publication date

  2019-12-01

• Fields of study

  Computer Science

• Identifiers
  DOI 10.1016/J.KNOSYS.2019.104909
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Virtual Subconcept Drift Detection in Discrete Data Using Probabilistic Graphical Models

  2018cited by this paper
• Dynamic financial distress prediction with concept drift based on time weighting combined with Adaboost support vector machine ensemble

  2017cited by this paper
• Bayesian Models of Data Streams with Hierarchical Power Priors

  2017cited by this paper
• A survey on data preprocessing for data stream mining: Current status and future directions

  2017cited by this paper
• An Overview of Concept Drift Applications

  2016cited by this paper
• Generating discrete analogues of continuous probability distributions-A survey of methods and constructions

  2015cited by this paper
• Learning from Data Streams in Dynamic Environments

  2015cited by this paper
• A survey on concept drift adaptation

  2014influential reference
• Dynamic financial distress prediction using instance selection for the disposal of concept drift

  2011cited by this paper
• Tracking recurring contexts using ensemble classifiers: an application to email filtering

  2009cited by this paper
• Knowledge discovery from data streams

  2008cited by this paper
• Learning from Time-Changing Data with Adaptive Windowing

  2007influential reference
• Anomaly detection in a mobile communication network

  2007cited by this paper
• Statistical Comparisons of Classifiers over Multiple Data Sets

  2006cited by this paper
• Some Variations on the PC Algorithm

  2006cited by this paper
• IMPROVING THE PRECISION OF ESTIMATES OF THE FREQUENCY OF RARE EVENTS

  2005cited by this paper
• Feature Based Modelling: A methodology for producing coherent, consistent, dynamically changing models of agents' competencies

  2005cited by this paper
• Detecting Change in Data Streams

  2004cited by this paper
• Learning with Drift Detection

  2004cited by this paper
• Learning drifting concepts: Example selection vs. example weighting

  2004cited by this paper
• Learning in the presence of concept drift and hidden contexts

  2004cited by this paper
• On-line Variational Bayesian Learning

  2003cited by this paper
• Models and issues in data stream systems

  2002cited by this paper
• Maintaining Stream Statistics over Sliding Windows

  2002cited by this paper
• Dynamic bayesian networks: representation, inference and learning

  2002cited by this paper
• Mining time-changing data streams

  2001cited by this paper
• Gradual Forgetting for Adaptation to Concept Drift

  2000cited by this paper
• A Tutorial on Learning with Bayesian Networks

  1999cited by this paper
• Probabilistic reasoning in intelligent systems

  1988cited by this paper

• Evolving Strategies in Machine Learning: A Systematic Review of Concept Drift Detection

  2024cites this paper
• Bayesian-based water leakage detection with a novel multisensor fusion method in a deep manned submersible

  2021cites this paper
• Dynamically adaptive and diverse dual ensemble learning approach for handling concept drift in data streams

  2021influential citation
• Remaining useful life re-prediction methodology based on Wiener process: Subsea Christmas tree system as a case study

  2020cites this paper
• Federated Learning for Personalized Humor Recognition

  2020cites this paper