English
 
Help Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT

Released

Talk

Mathematical Programming Methods for Learning from Structured Data: Applications to Molecular QSAR Analysis, HIV Drug Resistance Prediction, and Image Categorisation

MPS-Authors
/persons/resource/persons84265

Tsuda,  K
Department Empirical Inference, Max Planck Institute for Biological Cybernetics, Max Planck Society;
Max Planck Institute for Biological Cybernetics, Max Planck Society;

/persons/resource/persons84183

Saigo,  H
Department Empirical Inference, Max Planck Institute for Biological Cybernetics, Max Planck Society;
Max Planck Institute for Biological Cybernetics, Max Planck Society;

/persons/resource/persons84113

Nowozin,  S
Department Empirical Inference, Max Planck Institute for Biological Cybernetics, Max Planck Society;
Max Planck Institute for Biological Cybernetics, Max Planck Society;

Fulltext (restricted access)
There are currently no full texts shared for your IP range.
Fulltext (public)
There are no public fulltexts stored in PuRe
Supplementary Material (public)
There is no public supplementary material available
Citation

Tsuda, K., Saigo, H., Nowozin, S., Kudo, T., & Uno, T. (2007). Mathematical Programming Methods for Learning from Structured Data: Applications to Molecular QSAR Analysis, HIV Drug Resistance Prediction, and Image Categorisation. Talk presented at Joint Bioinformatics Education Program: Kyoto University and University of Tokyo. Kyoto, Japan. 2007-05-14.


Cite as: https://hdl.handle.net/21.11116/0000-0004-4A31-2
Abstract
In learning from structured data such as graphs, trees and itemsets, frequent substructure mining methods are used for deriving features. However, frequent patterns are not necessarily informative for the given learning problem. We propose a mathematical programming boosting method that progressively collects informative patterns. Compared to AdaBoost, our method can build the prediction rule with fewer iterations. To apply the boosting method to graph data, a branch-and-bound-type pattern search algorithm is developed based on the DFS code tree. The constructed search space is reused in later iterations to minimize the computation time. Our method can learn more efficiently than the simpler method based on frequent substructure mining, because the output labels are used as an extra information source for pruning the search space. Furthermore, by engineering the mathematical program, a wide range of machine learning problems can be solved without modifying the pattern search algorithm. I will show practical applications to graphs (QSAR) and itemsets (HIV and images).