a tutorial on support vector machines for pattern recognition asli taŞÇi christopher j.c. burges,...
TRANSCRIPT
A TUTORİAL ON SUPPORT VECTOR MACHİNES FOR PATTERN RECOGNİTİON
ASLI TAŞÇİ
Christopher J.C. Burges, Data Mining and
Knowledge Discovery 2, 121-167, 1998
OUTLİNE
• Introduction• Linear Support Vector Machines• Nonlinear Support Vector Machines• Limitations• Conclusion
INTRODUCTİON
Classification and Regression tool
Supervised Learning
Linear and non-linear classification performance
APPLİCATİON AREAS
Handwritten Digit Recognition
Object Recognition
Speaker Identification
Text Categorization
Face Detection in Images
LİNEAR SUPPORT VECTOR MACHİNES
Simplest Case: Seperable Data
SVM Equaiton:
Lagranian:
KARUSH-KUHN-TUCKER CONDİTİONS
Constraint optimization
NON-SEPERABLE CASEIntroducing Slack variables for a feasible solution with linear SVM
Lagranian for non-seperable data:
NONLİNEAR SUPPORT VECTOR MACHİNES
Mapping data to a feature space
Example:
Kernel Function:
MERCER’S CONDİTİON
Positive Semi-definite
OPTİMİZATİON PROBLEM
Quadratic programming optimizaiton
TRAİNİNG
Decomposition algorithms for larger problems• Chunking method• Osuna’s decomposition algorithm
LİMİTATİONS
• Choice of the Kernel• Speed• Size• Discrete Data• Multi-class classification
PERFORMANCE OF SVMThe Virtual Support Vector Method• Training the system than creating a
new data by distorting the resulting support vectors.
The reduced set method• Increases the speed of SVM
CONCLUSİON
• New approach to the problem of pattern recognition
• SVM training always find a global minimum• Largely characterized by the choice of its
Kernel
THANK YOU FOR LİSTENİNG