Instance-Based Learning Algorithms

By David William Aha, Dennis F. Kibler, Marc K. Albert. 1991

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

Black-Box vs. White-Box: Understanding Their Advantages and Weaknesses From a Practical Point of View (2019, methodology): “a lazy learning) where the function is only approximated locally, like the k-nearest neighbors [20], [21].”
Graph-based approach for airborne light detection and ranging segmentation (2014, methodology): “IBk w (Aha et al., 1991) is a KNN classifier which tunes K using cross-validation with linear neighbor search and Euclidean distance.”
A review of instance selection methods (2010, methodology): “…Misclassification Wilson 1972 All k-NN W Misclassification Tomek 1976 Multiedit W Misclassification Devijver and Kittler 1980 IB W Misclassification Aha et al. 1991 DROP W Associates Wilson and Martínez 2000 ICF W Reachable, Coverage Brighton and Mellish 2002 SV-kNNC W SVM, k-means Srisawat et al.…”
Ensemble learning for data stream analysis: A survey (2017, result): “This approach was shown to achieve comparable or better performance than previous approaches such as FLORA4 [184] and instance-based learning 3 (IB3) [3] in the presence of concept drifts, but sometimes presented worse performance than FLORA4 before the drifts.”
Advances in Instance Selection for Instance-Based Learning Algorithms (2002, methodology): “1974; Zhang, 1992, Sebban et al., 1999) or instances with high utility (Markovitch and Scott, 1988, 1993; Smyth and Keane, 1995; Aha et al., 1991 ).”
Dynamic weighted majority: a new ensemble method for tracking concept drift (2003, background): “Although removing experts of low weight yielded positive results for the problems we considered in this study, it would be benecial to investigate mechanisms for explicitly handling noise, such as those present in ib3 [ 15 ], or for determining when examples are likely to be from a dieren t target c”
Ontological user profiling in recommender systems (2004, methodology): “Research papers in the central database are classi.ed by an IBk [Aha et al. 1991] classi.er, which is boosted by the AdaBoostM1 [Freund and Schapire 1996] algorithm.”
Classifier Ensembles for Detecting Concept Change in Streaming Data: Overview and Perspectives (2008, methodology): “In the classifier-specific case, the forgetting mechanism can only be applied to a chosen and fixed classifier model [1,3] or classifier ensemble [30].”
Building Acceptable Classification Models (2010, methodology): “In the literature, there is a myriad of different techniques proposed for this classification task [4, 22], some of the most commonly used being C4.5 [42], CART [10], logistic regression [24], linear and quadratic discriminant analysis [9, 24], k-nearest neighbor [ 1 , 18, 24], artificial neural net”
Learning in the Presence of Concept Drift and Hidden Contexts (1996, methodology): “Simple Instance-Based Learning (sometimes called memory-based learning) algorithms like IB1 (Aha et al., 1991) can be viewed as incremental on-line learners that rst classify each newly arrived example by some nearest-neighbor method and then store it as a new exemplar.”
Reducing the effort of bug report triage: Recommenders for development-oriented decisions (2011, methodology): “… [Gunn 1998], C4.5 [Quinlan 1993], Expectation Maximization [Dempster et al. 1977], conjunctive rules [Witten and Frank 2000], and nearest neighbour [Aha et al. 1991].12 We chose these algorithms as they cover the different categories of supervised machine learning algorithms [Witten and Frank…”