Title :
Adaptive Local Hyperplane for regression tasks
Author :
Kecman, Vojislav ; Yang, Tao
Author_Institution :
Comput. Sci. Dept., Virginia Commonwealth Univ., Richmond, NJ, USA
Abstract :
The paper introduces novel machine learning (data mining) algorithm called Adaptive Local Hyperplane (ALH) and it presents its application in solving regression problems. ALH algorithm has recently shown extremely good results in classification, and it is adopted for solving regression tasks here. It is a local margin maximizing algorithm in the original, weighted, input space blending a Nearest Neighbors (NN) based approaches and Support Vector Machines (SVMs) ideas about the maximal margin. In performing such a task it uses only K closest points to the query data point. Results for four benchmarking regression data sets show superior performance to SVMs as well as to the other established regression methods.
Keywords :
data mining; learning (artificial intelligence); mathematics computing; optimisation; pattern classification; regression analysis; support vector machines; K closest point; adaptive local hyperplane; classification task; data mining; local margin maximizing algorithm; machine learning algorithm; nearest neighbor; regression task; support vector machine; Classification tree analysis; Data mining; Face recognition; Function approximation; Machine learning; Machine learning algorithms; Nearest neighbor searches; Neural networks; Support vector machine classification; Support vector machines;
Conference_Titel :
Neural Networks, 2009. IJCNN 2009. International Joint Conference on
Conference_Location :
Atlanta, GA
Print_ISBN :
978-1-4244-3548-7
Electronic_ISBN :
1098-7576
DOI :
10.1109/IJCNN.2009.5178919
