Communications of the ACM - Special issue on parallelism
Radial basis functions for multivariable interpolation: a review
Algorithms for approximation
Instance-Based Learning Algorithms
Machine Learning
Tolerating noisy, irrelevant and novel attributes in instance-based learning algorithms
International Journal of Man-Machine Studies - Special issue: symbolic problem solving in noisy and novel task environments
Similarity metric learning for a variable-kernel classifier
Neural Computation
Unifying instance-based and rule-based induction
Machine Learning
Reduction Techniques for Instance-BasedLearning Algorithms
Machine Learning
Selecting Typical Instances in Instance-Based Learning
ML '92 Proceedings of the Ninth International Workshop on Machine Learning
ICML '97 Proceedings of the Fourteenth International Conference on Machine Learning
ICML '97 Proceedings of the Fourteenth International Conference on Machine Learning
Improved heterogeneous distance functions
Journal of Artificial Intelligence Research
Rule induction and instance-based learning a unified approach
IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 2
Multiresolution instance-based learning
IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 2
Symbolic nearest mean classifiers
AAAI'97/IAAI'97 Proceedings of the fourteenth national conference on artificial intelligence and ninth conference on Innovative applications of artificial intelligence
A novel Supervised Instance Selection algorithm
International Journal of Business Intelligence and Data Mining
A method of learning weighted similarity function to improve the performance of nearest neighbor
Information Sciences: an International Journal
A proposed method of local feature-weighting to improve predictions of basic nearest neighbor rule
ASC '07 Proceedings of The Eleventh IASTED International Conference on Artificial Intelligence and Soft Computing
InstanceRank: Bringing order to datasets
Pattern Recognition Letters
Automated constraint selection for semi-supervised clustering algorithm
CAEPIA'09 Proceedings of the Current topics in artificial intelligence, and 13th conference on Spanish association for artificial intelligence
PolarityRank: Finding an equilibrium between followers and contraries in a network
Information Processing and Management: an International Journal
Hi-index | 0.00 |
Two disadvantages of the standard nearest neighbor algorithm are1) it must store all the instances of the training set, thuscreating a large memory footprint and 2) it must search all theinstances of the training set to predict the classification of anew query point, thus it is slow at run time. Much work has beendone to remedy these shortcomings. This paper presents a newalgorithm WITS (Weighted-Instance Typicality Search) and a modifiedversion, Clustered-WITS (C-WITS), designed to address these issues.Data reduction algorithms address both issues by storing and usingonly a portion of the available instances. WITS is an incrementaldata reduction algorithm with O(n^2) complexity, where n is thetraining set size. WITS uses the concept of Typicality inconjunction with Instance-Weighting to produce minimal nearestneighbor solutions. WITS and C-WITS are compared to three otherstate of the art data reduction algorithms on ten real-worlddatasets. WITS achieved the highest average accuracy, showed fewercatastrophic failures, and stored an average of 71% fewer instancesthan DROP-5, the next most competitive algorithm in terms ofaccuracy and catastrophic failures. The C-WITS algorithm provides auser-defined parameter that gives the user control over thetraining-time vs. accuracy balance. This modification makes C-WITSmore suitable for large problems, the very problems data reductionsalgorithms are designed for. On two large problems (10,992 and20,000 instances), C-WITS stores only a small fraction of theinstances (0.88% and 1.95% of the training data)while maintaininggeneralization accuracies comparable to the best accuraciesreported for these problems.