Classification Based on Combination of Kernel Density Estimators

Authors:
Mateusz Kobos;Jacek Mańdziuk
Affiliations:
Faculty of Mathematics and Information Science, Warsaw University of Technology, Warsaw, Poland 00-661;Faculty of Mathematics and Information Science, Warsaw University of Technology, Warsaw, Poland 00-661
Venue:
ICANN '09 Proceedings of the 19th International Conference on Artificial Neural Networks: Part II
Year:
2009

Citing 7
Cited 3

A Comparison of Prediction Accuracy, Complexity, and Training Time of Thirty-Three Old and New Classification Algorithms

Machine Learning
Pattern Recognition and Neural Networks

Pattern Recognition and Neural Networks
Linearly Combining Density Estimators via Stacking

Machine Learning
Looking for lumps: boosting and bagging for density estimation

Computational Statistics & Data Analysis - Nonlinear methods and data mining
Pattern Classification (2nd Edition)

Pattern Classification (2nd Edition)
Improving Performance of a Binary Classifier by Training Set Selection

ICANN '08 Proceedings of the 18th international conference on Artificial Neural Networks, Part I
Averaging, maximum penalized likelihood and Bayesian estimation for improving Gaussian mixture probability density estimates

IEEE Transactions on Neural Networks

Classification based on multiple-resolution data view

ICANN'10 Proceedings of the 20th international conference on Artificial neural networks: Part III
Bandwidth selection in kernel density estimators for multiple-resolution classification

ICAISC'12 Proceedings of the 11th international conference on Artificial Intelligence and Soft Computing - Volume Part I
A probabilistic approach to mining geospatial knowledge from social annotations

SIGSPATIAL Special

Quantified Score

Hi-index	0.00

Visualization

Abstract

A new classification algorithm based on combination of kernel density estimators is introduced. The method combines the estimators with different bandwidths what can be interpreted as looking at the data with different "resolutions" which, in turn, potentially gives the algorithm an insight into the structure of the data. The bandwidths are adjusted automatically to decrease the classification error. Results of the experiments using benchmark data sets show promising performance of the proposed approach when compared to classical algorithms.