Artificial Intelligence II (CS4442 & CS9542)

2025 Winter

Department of Computer Science

University of Western Ontario
Outline

▶ Administrative Issues

▶ Artificial Intelligence

▶ Machine Learning

▶ Deep Learning

1
Administrative Issues

2
Administrative Issues
▶ Lectures: Mondays (9:30-11:30am) and Wednesdays (9:30-10:30am).

▶ Instructors: Yalda Mohsenzadeh & Boyu Wang

- Emails: [email protected] & [email protected]

- Boyu Wang Office Hours: 10:30am-11:30am on Wednesdays @
MC-366 or by appointment
- Yalda Mohsenzadeh Office Hours: 10:30pm-11:30pm on
Wednesdays at MC-385 or by appointment
▶ TAs: Gezheng Xu, Ruizhi Pu, Ruiyi Fang, Pengcheng Xu

- Email: [email protected], [email protected], [email protected],

[email protected]
- Office Hours: 2pm-3pm, Monday (Gezheng Xu);
10am-11am, Tuesday (Ruiyi Fang);
1pm-2pm, Thursday (Ruizhi Pu);
10am-11am, Friday (Pengcheng Xu)
- Zoom link: https://westernuniversity.zoom.us/j/99212891950 (no 3
passcode)
Course Website: OWL

▶ Most announcements about the course, lecture materials, or the

assignments should be addressed via OWL.

▶ Long, detailed questions are probably best answered during

office hours.

▶ Use your judgement.

4
Class Materials

▶ No required textbook, but several textbooks available describing

parts of the material that we will cover.

- Bishop, Pattern Recognition and Machine Learning, Springer,

2006.
- Murphy, Machine Learning: A Probabilistic Perspective, MIT
Press, 2012.
- Goodfellow, Bengio, Courville, Deep Learning, MIT Press, 2016.
- Forsyth and Ponce, Computer Vision: A Modern Approach,
Prentice Hall, 2002.
- Individual papers or web resources may be assigned to
supplement lecture material.

▶ Class slides: posted on OWL.

5
Prerequisites

▶ Knowledge of a programming language (e.g., Python, Matlab)

▶ Solid knowledge of probability/statistics, calculus and linear

algebra; basic knowledge of optimization; general facility with
math

▶ Some AI background is recommended but not required

6
Evaluation – Undergraduate Students

▶ Four homework assignments (60%)

- Each assignment will involve components that are conceptual or

algorithmic, as well as some practical implementation.
- Each assignment must be submitted in PDF format. It is
recommended that assignments be completed in LATEX compiled
to PDF.
- You should also submit your code (e.g., .py, .ipynb, .m files)
- Homework assignments should be done individually.
- 10% of each assignment will be taken off each day for late
submissions; after 5 days being late, no points are given anymore.
- You should must obtain at least 50% score of the assignments to
pass the course.

▶ Group project (40%)

7
Evaluation – Graduate Students

▶ Four homework assignments (40%)

- Each assignment will involve components that are conceptual or

algorithmic, as well as some practical implementation.
- Each assignment must be submitted in PDF format. It is
recommended that assignments be completed in LATEX compiled
to PDF.
- You should also submit your code (e.g., .py, .ipynb, .m files)
- Homework assignments should be done individually.
- 10% of each assignment will be taken off each day for late
submissions; after 5 days being late, no points are given anymore.
- You should must obtain at least 50% score of the assignments to
pass the course.
▶ Group project (60%) – project presentation (10%) + project
report (50%)

8
Final Project

▶ Students are encouraged to formulate a project related to their

own interest or research

▶ Students who do not have such problems should contact Yalda

and/or Boyu to discuss possible projects

▶ Completed in groups of 2-4 members (undergraduate students)

or 2-3 members (graduate students)

▶ Both undergraduate and graduate students are required to write

a project report

▶ Graduate students are also required to do a final project

presentation (10 minutes per team). The presentations will be
scheduled in the last week of class, during of the class time.

9
Artificial Intelligence

10
What is Artificial Intelligence (AI)?

▶ Originally, AI is a subfield of computer science, aiming to build a

machine solving tasks that humans are good at (e.g., natural
language, speech, image recognition, ...)

▶ Interdisciplinary science with multiple approaches (computer

science, philosophy, psychology, neuroscience...)

▶ Most recent achievements are in machine learning, especially

deep learning.

11
AI & Machine Learning & Deep Learning

Deep Machine Artificial

Learning Learning Intelligence

Deep neural networks (with E.g., search, planning, logic

many layers and parameters) E.g., linear model, rules, symbolic expression,
SVM, “shallow” expert system
neural networks

12
▶ This course: introduction to artificial intelligence

13
▶ This course: introduction to artificial intelligence

13
▶ This course: introduction to machine learning

13
▶ This course: introduction to

- Cover (some of) the most commonly used machine learning

concepts and algorithms.

- Focus on the mechanisms of the algorithms: explain the

problem settings, the mathematical formulations, why they
work; but NOT how to implement them.

- Deep learning and its applications on computer vision

(Yalda Mohsenzadeh)

13
Machine Learning

14
What is Machine Learning?

▶ Machine learning is the field of study that gives computers the

ability to learn without being explicitly programmed
– Arthur Samuel

▶ Learning is any process by which a system improves

performance from experience
– Herbert Simon

Machine learning aims to construct a computer system that acts like

humans do by learning from data.

15
Machine Learning is Changing the World

▶ A breakthrough in machine learning would be worth ten

Microsofts
– Bill Gates, Microsoft

▶ Web rankings today are mostly a matter of machine learning

– Prabhakar Raghavan, Google

▶ Machine learning is the hot new thing

– John Hennessy, Stanford

16
Why (Study) Machine Learning?

▶ Easier to build a learning system than to hand-code a working

program (play games)

▶ Human expertise does not exist (navigating on Mars)

▶ Humans can’t explain their expertise (speech recognition)

▶ Models are based on huge amounts of data (DNA analysis)

▶ Models must be customized (news filtering)

▶ Discover knowledge and patterns in highly dimensional, complex

data (social network analysis)

▶ Understanding animal and human learning (language,

recognition)

▶ Create real AI! 17

Sample Applications

Medicine
Finance and economics

Computer vision

Software engineering

Machine Learning

Natural language
processing

Neuroscience and
neural engineering

18
Web applications
Smart grids and smart cities
Autonomous Car Technology

Images taken from Sebastian Thruns multimedia website.

19
Speech Technology

Slide credit: Li Deng, MS Research

20
Deep Learning in the Headlines

Slide credit: Eric Eaton, UPenn

21
Other Success Stories

[Krizhevsky et al. NIPS, 2012]

[Silver et al. Nature, 2016, 2017]

https://openai.com/blog/openai-five/

22
Types of Learning

▶ Supervised learning

- Given: data + desired outputs (labels)

▶ Unsupervised learning

- Given: data (without desired outputs)

▶ Semi-supervised learning

- Given: data + a few desired outputs

▶ Reinforcement learning

- Given: Rewards from sequence of actions (an interactive

process)

23
Supervised Learning: Regression

Given (x1 , y1 ), (x2 , y2 ), . . . , (xn , yn ), learn a function f (x) to predict y

given x. If y is real-valued ⇒ regression.

24
Supervised Learning: Classification

Given (x1 , y1 ), (x2 , y2 ), . . . , (xn , yn ), learn a function f (x) to predict y

given x. If y is categorical ⇒ classification.

x2 𝒚 = +𝟏
(Malignant)
Age

𝒚 = −𝟏
(Benign)
Tumor Size x1
25
Unsupervised Learning

Given x1 , x2 , . . . , xn (without labels), Output hidden structure behind

the x’s (e.g., clustering) Image credit: Eric Eaton, UPenn

26
Anticipated Lecture Topics

▶ Supervised learning

- A brief review of mathematical tools

- Linear regression
- Logistic regression
- Regularization
- Support vector machines
- Kernel methods
- Ensemble methods
- Bayesian learning
▶ Unsupervised learning

- Clustering
- Dimensionality reduction
▶ Deep learning and computer vision (Yalda Mohsenzadeh)
27