A landmarker selection algorithm based on correlation and efficiency criteria

Authors:
Daren Ler;Irena Koprinska;Sanjay Chawla
Affiliations:
School of Information Technologies, University of Sydney, NSW, Australia;School of Information Technologies, University of Sydney, NSW, Australia;School of Information Technologies, University of Sydney, NSW, Australia
Venue:
AI'04 Proceedings of the 17th Australian joint conference on Advances in Artificial Intelligence
Year:
2004

Citing 10
Cited 0

Technical Note: Selecting a Classification Method by Cross-Validation

Machine Learning
Characterizing the applicability of classification algorithms using meta-level learning

ECML-94 Proceedings of the European conference on machine learning on Machine Learning
Machine learning, neural and statistical classification

Machine learning, neural and statistical classification
Data mining: practical machine learning tools and techniques with Java implementations

Data mining: practical machine learning tools and techniques with Java implementations
A perspective view and survey of meta-learning

Artificial Intelligence Review
Ranking Learning Algorithms: Using IBL and Meta-Learning on Accuracy and Time Results

Machine Learning
God Doesn't Always Shave with Occam's Razor - Learning When and How to Prune

ECML '98 Proceedings of the 10th European Conference on Machine Learning
Meta-Learning by Landmarking Various Learning Algorithms

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
Feature Selection for Meta-learning

PAKDD '01 Proceedings of the 5th Pacific-Asia Conference on Knowledge Discovery and Data Mining
Introduction to the Special Issue on Meta-Learning

Machine Learning

Quantified Score

Hi-index	0.00

Visualization

Abstract

Landmarking is a recent and promising meta-learning strategy, which defines meta-features that are themselves efficient learning algorithms However, the choice of landmarkers is often made in an ad hoc manner In this paper, we propose a new perspective and set of criteria for landmarkers Based on the new criteria, we propose a landmarker generation algorithm, which generates a set of landmarkers that are each subsets of the algorithms being landmarked Our experiments show that the landmarkers formed, when used with linear regression are able to estimate the accuracy of a set of candidate algorithms well, while only utilising a small fraction of the computational cost required to evaluate those candidate algorithms via ten-fold cross-validation.