Learning drifting concepts: Example selection vs. example weighting

By Ralf Klinkenberg. 2004

498 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems

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 Information Systems

Citations in the corpus

Online Machine Learning in Big Data Streams (2018, background): “ changes [162, 73]. For sudden changes, algorithms that maintain a sliding window of the last seen instances perform better [228, 93, 145]. Another option is to include explicit forgetting mechanisms [142, 136, 76]. The most important distinction is whether changes are explicitly or implicitly detec”
Learning in Nonstationary Environments: A Survey (2015, background): “As suggested in [71], samples can also be weighted according to the classification accuracy/error computed on the last batch of supervised data.”
On evaluating stream learning algorithms (2012, unknown): “We focus on change detectors that monitor the evolution of learning performance, a common strategy for drift detection metrics (Widmer and Kubat, 1996; Street and Kim, 2001; Klinkenberg, 2004).”
Minimax Classification under Concept Drift with Multidimensional Adaptation and Performance Guarantees (2022, unknown): “…learning techniques account for a scalar rate of change by means of a carefully chosen parameter such as a learning rate (Kivinen et al., 2004; Orabona et al., 2008), forgetting factor (Klinkenberg, 2004; Pavlidis et al., 2011), or window size (Bifet & Gavalda, 2007; Nguyen et al., 2018).”
Incremental Clustering for the Classification of Concept-Drifting Data Streams (2008, background): “Examples of this group can be found in [5, 10, 19].”
Incremental Evolving Domain Adaptation (2016, methodology): “methods such as [18], [19], [20], [21], [22], [23], [24], [25] are”
Dynamic Weighted Majority: An Ensemble Method for Drifting Concepts (2007, background): “, 2005) and synthetic (e.g., Schlimmer and Granger, 1986; Widmer and Kubat, 1996; Maloof and Michalski, 2000; Hulten et al., 2001; Street and Kim, 2001; Kolter and Maloof, 2003; Klinkenberg, 2004; Maloof and Michalski, 2004; Gama et al., 2005; Kolter and Maloof, 2005; Scholz and Klinkenberg, 2005) d”
Applying lazy learning algorithms to tackle concept drift in spam filtering (2007, methodology): “In Klinkenberg (2004), Klinkenberg shows that instance weighting techniques handle concept drift less eﬀectively than analogous instance selection techniques, which is probably due to over ﬁtting the data.”
UNICORN: Runtime Provenance-Based Detector for Advanced Persistent Threats (2020, background): “U NICORN accounts for such changes in system behavior through the use of exponential weight decay [70] on histogram element counts to gradually forget outdated data.”
Continual Learning with Gated Incremental Memories for sequential data processing (2020, background): “Similarly, instance weighting systems [11] exploit weights on the input patterns which can inform predictions based”
Online and Non-Parametric Drift Detection Methods Based on Hoeffding’s Bounds (2015, methodology): “To improve the detection on gradual and slow types of change, Baena et al. [4] consider the distance between two classification errors instead of considering only the number of errors.”
Tracking recurring contexts using ensemble classifiers: an application to email filtering (2009, methodology): “To achieve this goal, a weighting scheme is defined [21].”
Learning with Drift Detection (2004, methodology): “There are several methods in machine learning to deal with changing concepts [7, 6, 5 , 12].”