Variational Bayesian Approach to Support Vector Regression
スポンサーリンク
概要
- 論文の詳細を見る
We consider a variational Bayesian approach to support vector regression (SVR). Its main advantage is that one can estimate the leave-one-out error of an SVR analytically without doing the cross-validation. Comparing our theory with the simulations on both artificial (the "sinc" function) and benchmark (the "Boston Housing") datasets, we get a good agreement. Furthermore, the smoothness of the hyperparameter dependence of the leave-one-out error can be tuned, which is useful for determining the optimal hyperparameters.
- 理論物理学刊行会の論文
- 2005-04-30
著者
-
WONG K.
Department of Applied Physics and Center for Smart Materials, Hong Kong Polytechnic University
-
GAO Zhuo
Department of Physics, Hong Kong University of Science and Technology
-
Gao Zhuo
Department Of Physics Hong Kong University Of Science And Technology:department Of Physics Beijing N
-
Wong K.
Department Of Applied Physics And Center For Smart Materials Hong Kong Polytechnic University
-
GAO Zhuo
Department of Physics, Hong Kong University of Science and Technology:Department of Physics, Beijing Normal University
関連論文
- Effects of Sr_Ba_Nb_2O_6 Self-Template Layer on the Structural Properties of Sol-Gel Derived Sr_Ba_Nb_2O_6 Films
- Cooperating Agents in the Minority Game
- Simultaneous determination of unbound ropivacaine in rat blood and brain using microdialysis
- Learning Dynamics as a Many-Body Problem (多体問題としての情報処理--統計力学と情報科学の接点)
- Variational Bayesian Approach to Support Vector Regression
- Laparoscopic hernioplasties in children : the implication on contralateral groin exploration for unilateral inguinal hernias
- Frequent DAP kinase but not p14 or Apaf-1 hypermethylation in B-cell chronic lymphocytic leukemia
- Recurrence After a First Untreated Seizure in the Hong Kong Chinese Population
- Effects of Sr0.6Ba0.4Nb2O6 Self-Template Layer on the Structural Properties of Sol-Gel Derived Sr0.6Ba0.4Nb2O6 Films
- 27pXZC-6 Entropy landscape analysis of the binary perceptron problem