CpSc 881: Machine Learning
Introduction
2
Copy Right Notice
Most slides in this presentation areadopted from slides of text book andvarious sources. The Copyright belong tothe original authors. Thanks!
3
General Information
Class Time: 5:45 PM ~ 8:30PM Monday
Location: 119 McAdams
Instructor:Dr. Feng Luo
Office:210 McAdams Hall
Phone:864-656-4793
Email:luofeng@clemson.edu
Office Hours:4:30PM ~ 5:30PM Monday
Web site:
http://www.cs.clemson.edu/~luofeng/course/MachineLearning/machinelearning.html
4
Prerequisite
Familiarity with basic computerscience principles and skills.
Familiarity with the basicmathematics, like probabilitytheory, basic linear algebra.
5
Text Book
Tom Mitchell. Machine Learning, 1997.ISBN 0-07-042807-7, WCB/McGraw-Hill
Reference books:
Ethem Alpaydin. Introduction to MachineLearning, 2004. ISBN: 0-262-01211-1, the MITPress.
Nils J. Nilsson Introduction to MachineLearning,(http://robotics.stanford.edu/people/nilsson/mlbook.html)
6
Grading
Grading:
Mid-term exam25 %
Final exam 25 %
Term project50 %
Curved to A, B, C,D
7
Resources: Datasets
8
Tools
Weka(http://www.cs.waikato.ac.nz/ml/weka/)
R (http://www.r-project.org/)
Octave: A free matlab clone(http://www.gnu.org/software/octave/)
Machine Learning Tools in Java(http://sourceforge.net/projects/mldev/)
9
Resources: Journals
Machine Learning
Neural Computation
Neural Networks
IEEE Transactions on Neural Networks
IEEE Transactions on Pattern Analysis andMachine Intelligence
Annals of Statistics
Journal of the American Statistical Association
10
Resources: Conferences
International Conference on Machine Learning (ICML)
European Conference on Machine Learning (ECML)
Neural Information Processing Systems (NIPS)
Uncertainty in Artificial Intelligence (UAI)
Computational Learning Theory (COLT)
International Joint Conference on Artificial Intelligence (IJCAI)
International Conference on Neural Networks (Europe)
ICANN05: http://www.ibspan.waw.pl/ICANN-2005/
11
What is Machine Learning
Definition – Computing Dictionary:
The ability of a machine to improve its performance based onprevious results.
12
Why Machine Learning
Human expertise does not exist (navigatingon Mars)
Humans are unable to explain theirexpertise (speech recognition)
Solution changes in time (routing on acomputer network)
Solution needs to be adapted to particularcases (user biometrics)
13
Three niches for machine learning
Data Mining: using historical data toimprove decisions
Medical records -> medical knowledge
Software application we can NOT programby hand
Autonomous driving
Speech recognition
Self customizing programs
News reader that learns user interests
14
Applications of MachineLearning
Information filtering/classification
15
Applications of MachineLearning
Playing games
16
Applications of MachineLearning
Robotics
17
Applications of MachineLearning
fault detection/monitoringtechnical systems
18
Applications of MachineLearning
bioinformatics
19
Applications of MachineLearning
image classificationpicture processing
20
Applications of MachineLearning
Text/languageprocessing,classification,visualization, ...
21
Applications of MachineLearning
22
Connections of MachineLearning
ML
AI
Neurobiology
Control
Statistics
Optimization
Information theory
Psychology
Philosophy