Probabilistic Output of Support Vector Machines

Probabilistic Output of Support Vector Machines

劉子榮(Tzu-Jung Liu)

指導教授 ： 林智仁

國立臺灣大學/工學院/工業工程學研究所/碩士(2004年)

https://doi.org/10.6342/NTU.2004.01037

摘要

無資料

關鍵字

機率輸出

並列摘要

Support vector machine (SVM) is a promising
technique for data classification
and regression.
However, it provides only decision values but not posterior probability
estimates.
As many applications require probability outputs,
it is essential to study how to transform SVM outputs to probability values.
In this thesis, we study and compare various methods.

並列關鍵字

SVM

參考文獻

[1] R. R. Bailey, E. J. Pettit, R. T. Boro cho , M. T. Manry, and X. Jiang. Automatic recognition of usgs land use/cover categories using statistical and neural networks classi ers. In SPIE OE/Aerospace and Remote Sensing , Bellingham, WA, 1993. SPIE.

[3] G. W. Brier. Veri cation of forecasts expressed in probabilities. Monthly Weather Review, 78:1–3, 1950.

[5] C.-C. Chang and C.-J. Lin. IJCNN 2001 challenge: Generalization ability and text deco ding. In Proceedings of IJCNN. IEEE, 2001.

[7] C.-J. Lin. Formulations of supp ort vector machines: a note from an optimization p oint of view. Neural Computation, 13(2):307–317, 2001.

[9] S. G. Nash and A. Sofer. Linear and Nonlinear Programming. McGraw-Hill, 1996.