Info from 2007
News
18/11: Assignment 2 and competition results available (see
bottom of this page).
All the lecture slides (.ppt) in a single zip
file.
Solutions for Q2 and Q10b of sample exam now online (bottom of this page).
Solutions for tutorials 6,7,8,9,10,12 now online (in comments below). Also
partial solutions for sample exam (below) and for tutorials 2, 3 and 4 (in
hints below). 8/11: Tutorial 10 solutions corrected (Q1 and Q3). 9/11: Tutorial
8 (Q2), Tutorial 9 (Q1) and Tutorial 12 (Q1) corrected (numerical errors).
Corrected versions of slides for lectures 9,10 and 11 now online (links below).
See p14 of 9_2.pdf (XOR output unit bias had wrong sign), p10,16,17 (10_2.pdf)
and p1 (11_2.pdf) (weight update rule contained erroneous simplifications).
Rationale
This course describes and
discusses several algorithms and techniques within the fields of artificial
intelligence and machine learning that have found theoretical or practical
applicability in software design and engineering. Theoretical and practical
understanding in these areas equips the student with the insights and tools
required for solving complex and difficult problems, and for implementing them
in software. Specific topics include problem solving by search, knowledge
representation and inference, probabilistic reasoning, machine learning and
information retrieval.
This course is aimed at
students with a computer science/engineering background, with an interest in
data structures and computing algorithms, and an aptitude for realising
theoretical ideas in software.
Links
Material
Russell S. and Norvig P., Artificial Intelligence: A modern approach,
2nd ed., 2003. Prentice Hall.
Lecture slides are provided
in a 6-per-page or 2-per-page format (links in the Teaching plan below).
Teaching
plan
|
Week Number |
Monday's Date |
Lecture Number |
Lecture Thu
10-11.50am 47A-141 |
Russell
and Norvig, 2003 |
Tutorial Session Tue
12-1.50pm 47A-141 |
Assessment |
|
1 |
23 July |
1 |
Introduction to artificial intelligence, an agent-based
perspective (6pp, 2pp)
|
Chapters 1, 2 and 26: pp. 947-949, 958-960. |
No tutorial |
|
|
2 |
||||||
|
2 |
30 July |
3 |
Chapter 3 |
|
||
|
4 |
||||||
|
3 |
6 August |
5 |
Chapter 4 (except 4.4 and 4.5) |
|
||
|
6 |
||||||
|
4 |
13 August |
7 |
Chapter 6 |
|||
|
8 |
||||||
|
5 |
20 August |
9 |
|
|
||
|
10 |
||||||
|
6 |
27 August |
11 |
Chapter 13 + 7.1-7.2 |
Assignment 1 preparation |
|
|
|
12 |
||||||
|
7 |
3 September |
13 |
Principles of machine learning, decision trees (6pp, 2pp, examples with decision tree learning) |
Chapters 18 and 19 |
|
|
|
14 |
||||||
|
8 |
10 September |
15 |
Chapter 20 (20.1-20.2) |
Assignment 1 deadline (Monday 10th September, 5pm) |
||
|
16 |
||||||
|
9 |
17 September |
17 |
Neural networks (6pp, 2pp (corrected 31/10/07), more material on NNs, NetTalk audio (8MB)) |
Chapter 20 (20.5-onwards) |
|
|
|
18 |
||||||
|
|
24 September |
Mid-semester break (one week) |
||||
|
10 |
1 October |
19 |
Learning algorithms for neural networks (6pp, 2pp,
(corrected 30/10/07) more on NN learning) |
Chapter 20 (20.5-onwards) |
||
|
20 |
||||||
|
11 |
8 October |
21 |
Applying machine learning and discussion of assignment 2 (6pp, 2pp,(corrected
30/10/07) more on applications of NN) |
|
|
|
|
22 |
||||||
|
12 |
15 October |
23 |
Probabilistic language processing and information
retrieval (6pp, 2pp) |
Chapter 23 (not 23.3-4) |
Assignment 2 preparation |
|
|
24 |
||||||
|
13 |
22 October |
25 |
Review, ML/assignment 2 |
|
Assignment 2 deadline (Friday 26 Oct, 5pm. Grace period until 11pm Fri. 2nd Nov.) |
|
|
26 |
||||||
|
|
29 October |
Revision Period |
||||
|
Exam Week 1 |
5 November |
|
|
|
|
Final Exam |
|
Exam Week 2 |
12 November |
|
|
|
|
|
Assessment
COMP3702 will be assessed by
a final exam and assignments. Your final grade (on a 1 to 7 scale) will be
determined by combining the marks from the assessment components below.
- Final examination (60 marks)
- Two assignments (30 marks)
- Ten tutorials (10 marks)
COMP7702 will be assessed
by a final exam and assignments. Your final grade (on a 1 to 7 scale) will be
determined by combining the marks from the assessment components below.
- Final examination (60 marks)
- Two assignments (30 marks)
- Ten tutorials (10 marks)
The examination papers for
COMP3702 and COMP7702 are different. Assessment is described in detail in the
course profile.
A set of sample exam questions are available here.
Hints and partial solutions to the sample
exam. Q2
solution, Q10b
solution.
Results
Results available online here.
Competition Results
The competition for Assignment
2 involved testing on 100 characters drawn by the lecturer with a mouse, chosen
at random (uniform distribution) from A-Z. An extra mark for assignment 2 was
given to each entry with 94 or more correct class predictions on the test set.
The competition was won by Kevin Wu Won with his classifier correctly
identifying 96/100 letters. Equal second place with 94/100 was claimed by 4
groups, which included: Joshua Bartlett, Joshua Fox, Michael Muthukrishna,
Andrew Schrauf, Chong Zhe Wei and Li Jie. The results for all other students
are shown below. Unfortunately, we could not produce a result for some entries,
but overall, the quality of entries was very high. Thanks to all students for
their participation.
|
StudentID |
Competition Result (number of correct predictions out of 100) |
|
40057514 |
93 |
|
40111591 |
79 |
|
40594617 |
93 |
|
40740272 |
91 |
|
40742267 |
93 |
|
40742557 |
91 |
|
40755100 |
93 |
|
40763646 |
91 |
|
40766357 |
86 |
|
40767783 |
90 |
|
40772040 |
91 |
|
40784584 |
3 |
|
40790110 |
66 |
|
40790893 |
66 |
|
40802590 |
82 |
|
40944034 |
93 |
|
40976370 |
43 |
|
40977470 |
89 |
|
40982401 |
86 |
|
40996613 |
84 |
|
41000843 |
84 |
|
41194199 |
3 |
|
41245112 |
86 |
|
41284465 |
91 |
|
41321696 |
91 |
|
41325694 |
89 |
|
41357305 |
86 |
|
41361115 |
89 |
|
41368392 |
91 |
|
41486586 |
85 |