Вы находитесь на странице: 1из 20

Support Vector

Machine
(Classification)
DuongTB2

9/24/15

Application
Spam email filtering

9/24/15

Application
optical character recognition

9/24/15

Human Detection

9/24/15

Basic Idea
Define the margin of a linear classifier as the
width that the boundary could be increased by
before hitting a datapoint.

9/24/15

Basic Idea

Maximizing the
margin is good
according to
intuition.
f(x,w,b) = sign(w x +
b)

9/24/15

Non-linear SVM

Boosting
Kernel
method

9/24/15

Boosting

9/24/15

Boosting

9/24/15

Boosting

9/24/15

Kernel Method

9/24/15

The Kernel trick


The Linear classifier relies on dot product between
vectors

=> the dot product in high dimensional space becomes:

9/24/15

Dual Problem
)
Subject to (
Apply dual theorem, we have:

Subject to

9/24/15

Solve quadratic programming


problem
Gradient Descent

9/24/15

References
Christopher J.C. Burges. A tutorial on support vector machines for pattern
recognition. Data Mining and Knowledge Discovery, pages 121167, 1998.
John C.Platt. Fast training of support vector machine using sequential
minimal optimization. Advances in Kernel Method-Support Vector Learning,
pages 185208. MIT Press, 1999.
Yoav Freund and Robert E. Schapire. A Decision-Theoretic Generalization of
On-Line Learning and an Application to Boosting. Journal of computer and
system sciences, pages 119-139 , 1997.

9/24/15

THANK YOU

9/24/15

9/24/15

9/24/15

9/24/15

Sequential Minimal Optimization

9/24/15

Вам также может понравиться