Detecting Concept Drift with Support Vector Machines

By Ralf Klinkenberg, Thorsten Joachims. International Conference on Machine Learning, 2000

511 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/2b9ce3415654bd4d5f6c311adb26c1d50dcc1372

How researchers see this paper

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

Citations in the corpus

Transfer learning using computational intelligence: A survey (2015, background): “Concept drift refers to a scenario in which data arrives sequentially with changing distribution, and the goal is to predict the next batch given the previously-arrived data [10].”
Early Drift Detection Method (2005, background): “Other approach to detect concept drift monitors the values of three performance indicators [7]: accuracy, recall and precision over time.”
Dynamic Weighted Majority: An Ensemble Method for Drifting Concepts (2007, methodology): “This type of concept drift has been called virtual concept drift, as opposed to real concept drift (Klinkenberg and Joachims, 2000).”
Learning from Time-Changing Data with Adaptive Windowing (2007, methodology): “In particular, [13] checks all subwindows of the current window, like our first algorithm ADWIN does, and is specifically tailored to work with SVMs.”
Estimating the Generalization Performance of an SVM Efficiently (2000, background): “They can eeectively select between diierent preprocessing steps e.g. stemming or no stemming, kernel parameters , and good values for C. Similarly, they can be used to detect concept drift Klinkenberg & Joachims, 2000.”
Cross-domain video concept detection using adaptive svms (2007, methodology): “Examples in this category are the approach proposed by Klinkenberg and Joachims [6] and by Delany et al.”
Classifier Technology and the Illusion of Progress (2006, unknown): “The optimal cost/benefit tradeoffs and the target class priors seldom are known precisely, and often are subject to change (Zahavi and Levin [47]; Friedman and Wyatt [8]; Klinkenberg and Thorsten [29]).”
Adaptive Deep Models for Incremental Learning: Considering Capacity Scalability and Sustainability (2019, methodology): “These methods can be divided into three categories: sliding window based approaches, which maintain the nearest data items and discard old items [21]; evolving based approaches, which downweight previous data items according to time series [20]; ensemble based approaches, which can adaptively add…”
Ensemble learning for data stream analysis: A survey (2017, unknown): “This approach has been shown to perform well in comparison to previous approaches such as adaptive window size [97] and batch selection [98, 96].”
Applying lazy learning algorithms to tackle concept drift in spam filtering (2007, background): “Instance weighting uses the ability of some learning algorithms such as support vector machines to process weighted instances (Klinkenberg & Joachims, 2000; Syed, Liu, & Sung, 1999; Taylor, Nakhaeizadeh, & Lanquillon, 1997).”
Learning drifting concepts: Example selection vs. example weighting (2004, unknown): “– Window of “Adaptive Size” : The window adjustment algorithm described in Section 4.1 (and in [13]) adapts the window size to the current concept drift situation.”
Consensus-Based Distributed Support Vector Machines (2010, methodology): “This feature also allows adapting MoM-DSVM to jointly operate with algorithms that account for concept drift (Klinkenberg and Joachims, 2000).”
A framework for generating data to simulate changing environments (2007, background): “KEY WORDS Machine Learning, Changing environments, Concept drift, Population drift, Artificial data, Simulated data.”
Adaptive Anomaly Detection via Self-calibration and Dynamic Updating (2009, background): “tive time window on the training data [14], select representative training examples [13], or weigh the training examples [15].”
Online Machine Learning in Big Data Streams (2018, methodology): “nline decision trees based on a statistical change detector that works on sliding windows are described in [26, 24]. More examples include Bayesian models [94, 19], neural networks [95, 153], and SVM [215, 137]. Concept drift adaptation methods exist for clustering [199, 207]. Sliding window based d”
Incremental Support Vector Machine Learning: A Local Approach (2001, background): “The same holds for Joachims’ e y estimators [10, 11].”