Java程序辅导

05s1: COMP9417 Machine Learning and Data Mining
Course Introduction
March 3, 2005
Aims
As a result of successfully completing this course students will be able
to set up well-defined learning problems, apply effective algorithms to such
such problems and use the relevant theory to interpret and evaluate the
results.
By the end of the subject, students should be able to:
• set up a well-defined learning problem for a given task
• select and define a representation for data to be used as input to a
machine learning algorithm
• select and define a representation for the model to be output by a
machine learning algorithm
• compare algorithms according to the properties of their inputs and
outputs
COMP9417: March 3, 2005 Course Intro: Slide 1
• describe and develop algorithms in terms of the computational methods
used
• relate different algorithms in terms of similarities and differences in the
computational methods used
• express key concepts from the foundations of computational and
statistical learning theory and demonstrate their applicability
• use algorithms in applications to real-world data sets and collect results
to enable evaluation and comparison of their performance
COMP9417: March 3, 2005 Course Intro: Slide 2
Assumed knowledge/prerequisites
Note: change from previous years
Old – COMP3411 or COMP9414 Artificial Intelligence New – COMP9024
Data Structures and Algorithms or COMP2011 Data Organisation
Waivers granted where applicable.
In practical terms, some knowledge of basic statistics and logic will be
helpful, but not essential. Ability to program in some language, preferably
Java, is assumed.
COMP9417: March 3, 2005 Course Intro: Slide 3
Course Web Pages
http://www.cse.unsw.edu.au/~cs9417/
COMP9417: March 3, 2005 Course Intro: Slide 4
Staff
Staff Name Role Email Extension
Mike Bain Lecturer & mike@cse.unsw.edu.au 56935
Course Convenor
Plus one or two exciting guest lecturers, to be arranged.
COMP9417: March 3, 2005 Course Intro: Slide 5
Syllabus
A summary of the topics to be covered. More details will be available on
the course web pages as the course progresses.
Module 1: Fundamentals of machine learning and data mining.
Weeks 1-5. Introduction to and overview of machine learning and data
mining. Decision tree learning. Rule learning. Numerical prediction.
Instance-based learning. Genetic algorithms. Reinforcement learning.
Module 2: Computational and statistical foundations of machine learning.
Weeks 6-8. Computational learning theory. Probabilistic foundations and
methods. Evaluating hypotheses.
Module 3: Advanced Machine Learning Techniques.
Weeks 9-14. SVMs and Ensemble methods. Hidden Markov models.
Bayes classifiers. Unsupervised learning and Clustering. Logical and
Relational Learning.
COMP9417: March 3, 2005 Course Intro: Slide 6
Lectures and Labs
Lecture timetable
Day Time Location
Thursday 6-9pm Matthews B
Problem with the enrollment system means you had to enrol for
laboratories. However, labs will not run every week. Practical work
is designed to be done in your own time. But in order to provide help
for the assignments there will be labs arranged on an “as needed” basis
before the assignments are due.
COMP9417: March 3, 2005 Course Intro: Slide 7
Assignments
Assignments will involve the process of applying and modifying or
implementing machine learning software, using the tools and techniques
described in lectures. The first assignment will involve Weka, while the
second assignment will be a more open-ended machine learning application
project involving implementation of machine learning methods.
Assignment Description Due
1 Weka toolkit Week 5
2 Project Week 13
In keeping with requirements of the Academic Board regarding post-
graduate courses, post-graduates will complete a version of assignment 2
which will have additional requirements.
COMP9417: March 3, 2005 Course Intro: Slide 8
Plagiarism
All work submitted for assessment must be your own work. Assignments
must be completed individually. We regard copying of assignments, in
whole or part, as a very serious offence. We use sophisticated plagiarism
detection software to search for unreasonable similarities in submitted
work.
• Submission of work derived from another person, without their consent,
will result in automatic failure for the course with a mark of zero.
• Submission of work derived from another person with their knowledge,
or jointly written with someone else, will result in zero marks for the
submission.
• Allowing another student to copy from you will, at the very least, result
in a reduction in the mark awarded for your own assignment or lab
COMP9417: March 3, 2005 Course Intro: Slide 9
exercises. Do not provide your work to any other person, even people
who are not UNSW students. You will be held responsible for the
actions of anyone you provide your work to.
• Severe or second offences constitute academic misconduct, and will
result in automatic failure, or exclusion from the University.
COMP9417: March 3, 2005 Course Intro: Slide 10
Exams
There will be a one-hour mid-term exam and a two-and-a-half closed-book
written exams. The written exams contribute 55% of the overall mark for
the course.
Exam timetable
Exam Date
Mid-term Week 7
Final Exam period
In keeping with requirements of the Academic Board regarding post-
graduate courses, post-graduates will be required to obtain a pass in both
exams to pass the course.
COMP9417: March 3, 2005 Course Intro: Slide 11
Trial of extended version
This year we are trialling an extended version of the course.
Learning objectives - to introduce additional wider and deeper coverage
of topics in the area, thereby making them available to students who can
cover the full course content relatively quickly due to previous exposure
to core concepts and other aspects of the material.
Extended students will study more and may achieve bonus marks based
on that. Bonus marks will be available only in assignments 1 & 2.
Students will self-select for the extended version. This is how it will work:
• lectures will be accompanied with extra papers or links for downloading
• everyone is welcome to download and read these materials
• there will be additional questions available for bonus marks in
assignment 1
COMP9417: March 3, 2005 Course Intro: Slide 12
• there will be additional project topics available for bonus marks in
assignment 2
• these topics will be more demanding, and should be based on the
extended reading materials or equivalent alternatives
So it is quite simple: extended students are those who nominate to do the
extended versions of the assignments !
COMP9417: March 3, 2005 Course Intro: Slide 13
Assessment
Assessment Marks
assignment 1 15
mid-term exam 15
assignment 2 / project 30
final exam 40
NOTE: course mark is total of component marks !
COMP9417: March 3, 2005 Course Intro: Slide 14
Reference Books
Textbook:
Machine Learning, Tom Mitchell, (1997), McGraw-Hill
Reference books:
Data Mining*, Ian Witten and Eibe Frank, (2000), Morgan Kaufmann
Classification and Regression Trees, Breiman, Friedman, Olshen and Stone
(1984), Kluwer
C4:5: programs for Machine Learning, J. R. Quinlan (1993), Morgan
Kaufmann
Pattern Classification (2nd ed.), Duda, Hart and Stork, (2001), Wiley
Elements of Statistical Learning, Hastie, Tibshirani and Friedman, (2001),
Springer
Pattern Recognition and Neural Networks, Brian Ripley, (1996),
Cambridge
COMP9417: March 3, 2005 Course Intro: Slide 15
Software
* WEKA machine learning toolkit in Java
http://www.cs.waikato.ac.nz/ml/weka/
COMP9417: March 3, 2005 Course Intro: Slide 16