what is learning all about ? get knowledge of by study, experience, or being taught become aware...
Post on 21-Dec-2015
216 views
TRANSCRIPT
What is Learning All about ?What is Learning All about ?
Get knowledge of by study, experience, or being taught
Become aware by information or fromobservation
Commit to memory
Be informed of or receive instruction
A Possible Definition of LearningA Possible Definition of Learning
Things learn when they change their behavior in a way that makes them perform better in the future.
Have your shoes learned the shape of your foot ?
In learning the purpose is the learner’s, whereas
in training it is the teacher’s
Our Learning Tasks in the ClassOur Learning Tasks in the Class
Classification (Supervised learning) binary classification problem multi-class classification problem
Regression (Supervised learning)
Does your machine learn anything from you?
Who/which is the better teacher/algorithm?
The Mathematical Background The Mathematical Background Material in the ClassMaterial in the Class
Calculus (Multi-variable)
f (x1;x2;x3) = x21 + x2
2 + x23
What is the gradient of function
Linear Algebra
How to compute the distance between two parallel hyperplanes in ? Rn
eigenvalue, positive definite matrix, inner product, projection matrix etc.
Basic Concepts of Probability and Basic Concepts of Probability and StatisticsStatistics
Probability:
Statistics:
Random variables, probability distribution,expected value (mean), variance …
Confidence interval, testing hypothesis …
Classification ProblemClassification Problem2-Category Linearly Separable Case2-Category Linearly Separable Case
A-
A+
wx0w = í + 1
x0w = í à 1
x0w = í
Malignant
Benign
Support Vector MachinesSupport Vector MachinesMaximizing the Margin between Bounding Maximizing the Margin between Bounding
PlanesPlanes
x0w = í + 1
x0w = í à 1
A+
A-
jjwjj22
w
= Margin
Why Use Support Vector Machines (SVMs)?Why Use Support Vector Machines (SVMs)?Powerful tools for Data MiningPowerful tools for Data Mining
SVM classifier is an optimally defined surface
SVMs have a good geometric interpretation SVMs can be generated very efficiently Can be extended from linear to nonlinear case
Typically nonlinear in the input space Linear in a higher dimensional “feature” space Implicitly defined by a kernel function
Have a sound theoretical foundation Based on Statistical Learning Theory
Why We Maximize the Margin?(Based on Statistical Learning Theory)
The Structural Risk Minimization (SRM):
The expected risk will be less than or equal to
empirical risk (training error)+ VC (error) bound
íí w
íí
2 / VC bound
minVC bound , min21íí w
íí 2
2 , maxMargin
Two-spiral Dataset(94 White Dots & 94 Red Dots)