论文信息 - COMBINING NEURAL NETWORKS VOTING CLASSIFIERS AND ERROR CORRECTING OUTPUT CODES

COMBINING NEURAL NETWORKS VOTING CLASSIFIERS AND ERROR CORRECTING OUTPUT CODES

We show that error correcting output codes (ECOC) can further improve the eeects of error dependent adaptive resampling methods such as arc-lh. In traditional one-inn coding, the distance between two binary class labels is rather small, whereas ECOC are chosen to maximize this distance. We compare one-inn and ECOC on a multiclass data set using standard MLPs and bagging and arcing voting committees.

Kurt Hornik | Friedrich Leisch

[1] Shu Lin,et al. Error control coding : fundamentals and applications , 1983 .

[2] Kurt Hornik,et al. ARC-LH: A New Adaptive Resampling Algorithm for Improving ANN Classifiers , 1996, NIPS.

[3] Thomas G. Dietterich,et al. Error-Correcting Output Coding Corrects Bias and Variance , 1995, ICML.

[4] Thomas G. Dietterich,et al. Solving Multiclass Learning Problems via Error-Correcting Output Codes , 1994, J. Artif. Intell. Res..

[5] Yoav Freund,et al. A decision-theoretic generalization of on-line learning and an application to boosting , 1995, EuroCOLT.

[6] Leo Breiman,et al. Bagging Predictors , 1996, Machine Learning.

[7] Shigeki Miyake,et al. Bayes statistical behavior and valid generalization of pattern classifying neural networks , 1991, IEEE Trans. Neural Networks.

[8] Leo Breiman,et al. Bias, Variance , And Arcing Classifiers , 1996 .