文档介绍：INTR ODUCTION
TO
MA CHINE LEARNING
AN EARL Y DRAFT OF A PR OPOSED
TEXTBOOK
Nils J
Nilsson
Rob otics Lab oratory
Departmen t puter Science
Stanford Univ ersit y
Stanford
CA
e
mail
nilsson
cs
stanford
edu
Decem ber
c
Cop yrigh t
Nils J
Nilsson
This material ma y not b e copied
repro duced
or distributed without the
written p ermission of the cop yrigh t holder
Con ten ts
Preliminaries
In tro duction
What is Mac hine Learning
W ellsprings of Mac hine Learning
V arieties of Mac hine Learning
Learning Input
Output F unctions
T yp es of Learning
Input V ectors
Outputs
T raining Regimes
Noise
P erformance Ev aluation
Learning Requires Bias
Sample Applications
Sources
Bibliographical and Historical Remarks
Bo olean F unctions
Represen tation
Bo olean Algebra
Diagrammatic Represen tations
Classes of Bo olean F unctions
T erms and Clauses
DNF F unctions
i
CNF F unctions
Decision Lists
Symmetric and V oting F unctions
Linearly Separable F unctions
Summary
Bibliographical and Historical Remarks
Using V ersion Spaces for Learning
V ersion Spaces and Mistak e Bounds
V ersion Graphs