학술대회자료
Regularization based multi-output regression model tree
- 한국시뮬레이션학회
- 한국시뮬레이션학회 학술대회집
- 2017년 춘계학술대회 발표집
- 2017.04
- 4901 - 4905 (5 pages)
Multi-output regression has become an emerging problem in data-mining and machine learning. We propose a multi-output regression model tree to obtain both accuracy and interpretation. Each leaf of proposed model tree contains sparse linear models that exploit the relationship between response variables. We present an efficient splitting rule based on residual analysis. Experiments on several dataset identify the performance of the proposed model over benchmark methods.
1. Introduction
2. Method
3. Experiment