RMIT Classification: Trusted
Introduction
COSC 2673-2793 | Semester 1 2021 (Computational) Machine Learning
Image: Freepik.com
RMIT Classification: Trusted
Agenda
• Introduction – Teaching Team.
• Overview of the course.
• Foundations of ML.
By the end of the lecture, you will:
• Understand what the course is about and the assignment structure.
• Have a high-level understanding of what ML is.
COSC2673 | COSC2793 Week 1: Introduction 2
Teaching Team
RMIT Classification: Trusted
COSC2673 | COSC2793 Week 1: Introduction 3
RMIT Classification: Trusted
Teaching Team
Ø Course Coordinator & lecturer
• Dr. Ruwan Tennakoon
• Consultation Hours: Wednesday (weekly) 4.00 pm – 5.00 pm, MS Teams – see consultation channel.
Ø Tutors
• Andrew Chester
• Wei Qin Chuah
COSC2673 | COSC2793 Week 1: Introduction 4
RMIT Classification: Trusted
Dr Ruwan Tennakoon
Ruwan’s research interests: Computer Vision & Machine Learning Medical Image Analysis
Statistical Signal Processing
Emphysema Diagnosis
Prostate Cancer Detection
Retinal image Analysis
Medical Image Analysis
COSC2673 | COSC2793 Week 1: Introduction 5
RMIT Classification: Trusted
Motion Segmentation
3D-Scene Analysis
Automated inspection
COSC2673 | COSC2793 Week 1: Introduction 6
RMIT Classification: Trusted
Members of Teaching Team
Andrew Chester
Andrew’s PhD research integrates symbolic reasoning techniques and deep reinforcement
learning to try to create general purpose thinking agents.
Wei Qin Chuah
Wei’s PhD research focuses on developing robust deep learning models for stereo matching and depth estimation that can be applied to many applications such as autonomous driving, robotics, augmented reality and more.
COSC2673 | COSC2793 Week 1: Introduction 7
Overview of the course
RMIT Classification: Trusted
COSC2673 | COSC2793 Week 1: Introduction 8
RMIT Classification: Trusted
Course Introduction
Learn about machine learning (ML) conceptually and practically:
Ø Introduce basic machine learning concepts:
• Supervised, unsupervised and reinforcement learning techniques.
• Specific approaches such as deep neural networks.
Ø Learn how to apply ML techniques to a range of problems, using open-source Machine Learning toolkits.
Ø Learn how to analyse:
• What the algorithms are doing.
• Datasets & outputs from the applications.
Ø Assignments will involve a variety of real-world datasets from a variety of domains.
Ø Learn terminology and concepts so can learn other ML approaches not covered in course
Ø This course is focused on training skilled ML engineers and scientists, Not doing autoML (aims to reduce or eliminate the need for skilled computer scientists/engineers to build machine learning models).
COSC2673 | COSC2793 Week 1: Introduction 9
RMIT Classification: Trusted
Ultimate Judgement
The core challenge of ML is NOT: Ø Collecting Data
Ø Running algorithms
Ø Maximizing Performance
The core challenge is in:
Ø Deciding what data to use
Ø Deciding if an algorithm is suitable
Ø Deciding the most suitable performance measure
Ø Deciding which hypothesis is “the best” to use for a task
Ø Making an ultimate judgement of how to approximate the unknown target function
COSC2673 | COSC2793 Week1: Foundations of ML 10
RMIT Classification: Trusted
Ultimate Judgement
The core challenge is in analysis and evaluation.
This is the focus of the course
Running algorithms is necessary and important, but not the top priority
This balance is best though of as:
We are not looking for the “best hypothesis”
We are looking for the “best hypothesis for the task that you can justify”
COSC2673 | COSC2793 Week1: Foundations of ML 11
RMIT Classification: Trusted
Assumed Knowledge Official Pre-requisites (ML)
Ø COSC 2627 – Discrete structures in Computing or MATH 1150 – Discrete Mathematics
Ø COSC 1076 – Advanced Programming Techniques
Official Co-requisites
Ø COSC 2123 – Algorithms and Analysis
Desirable
ü Knowledge of Python
ü Calculus (gradients)
ü Linear Algebra
ü Optimization
ü Probability and Statistics ü Algorithms
Official Pre-requisites (Computational ML) Ø COSC 1285 – Algorithms and Analysis
COSC2673 | COSC2793 Week 1: Introduction 12
RMIT Classification: Trusted
Syllabus
Week Topic
1 Foundations of ML
2 Regression
3 Logistic Regression & Regularisation
4 Evaluating Hypothesis & Bayesian Learning
5 Decision Tree Learning
6 Rule Learning
7 Reinforcement Learning
8 Practical Issues & Dimensionality Reduction
9 Neural Networks
10 Deep Learning
11 Unsupervised Learning
12 TBA & Review
COSC2673 | COSC2793 Week 1: Introduction 13
RMIT Classification: Trusted
Expectations
Us → You
Ø This is an elective (for most of you). Hence we do expect you to work hard and be highly motivated
Ø Designed to imitate Face to Face, please attend classes.
Ø Interact while in class! Be curious, ask questions. In this class, we have the “no stupid
question” rule!
Ø Follow standards of Academic Integrity and Ethical Behaviour
You → Us
Ø We will strive to be prompt in replying to your requests
Ø We provide support and help for you to achieve the best learning outcomes as you can
Ø We try to be fair as possible and allow you opportunities to ask us about assessment marks but please do not haggle about them.
COSC2673 | COSC2793 Week 1: Introduction 14
RMIT Classification: Trusted
Course Delivery Structure
COSC2673 | COSC2793 Week 1: Introduction 15
RMIT Classification: Trusted
Course Structure
Assignments
Apply concepts to solve real problems
Pre-Recorded Video & Q&A
Learn main concepts and explore examples
Labs
Practice applying main concepts to solve problems
Tutorials
Practice and consolidate theoretical concepts
Mathematical & programming skills
COSC2673 | COSC2793 Week 1: Introduction 16
RMIT Classification: Trusted
Course Delivery & Structure
This semester COSC 2673/2793 will be delivered partially online through a combination
Each week there will be:
Ø Pre-recorded videos of the weekly course content. To be viewed before the Q&A session.
Ø Live Q&A session (during the lecture time slot on the timetable) which will feature back-and- forth discussion where we will put into practice the concepts discussed each week in the videos.
Ø Lab exercises that can be done at home. There are also weekly scheduled lab classes where you can get help from the tutors.
Ø Tutorials containing exercises to expand your theoretical understanding of ML. It is recommended to attempt the question yourself before attending the online/on-campus sessions.
Ø Self Study You are also expected to spend a significant amount of time in private study, working through the course as presented in classes and learning materials and gaining practice at solving conceptual and technical problems.
of written/recorded content and face-to-face online/on-campus classes.
COSC2673 | COSC2793 Week 1: Introduction 17
RMIT Classification: Trusted
Course Delivery & Structure
Classroom and Hours:
• Pre-Recorded Videos: Microsoft stream channel
• Lectures (Q&A sessions): Tuesday 17:30-18:30 Thursday 15:30 – 16:30. Via MSTeams.
• SeeWeek0Gettingstartedmoduleoncanvastojoin.
• Tutorials: ColaborateUltra – See Canvas
• Checktimetablefordates
• Laboratories: ColaborateUltra – See Canvas
• Checktimetablefordates
Canvas:
Everything (slides/homeworks/projects/discussion board etc.)
COSC2673 | COSC2793 Week 1: Introduction 18
RMIT Classification: Trusted
Getting Started Checklist
If you haven’t done so yet, go through the checklist below to get started
Ø Admin activity
ü Join the course MS Team: Follow instructions at MS Team Group for COSC
2673/2793.
ü Make sure you can view the MS streams channel for weekly videos: Click “Weekly videos” on the navigation panel (There is some delay between joining the MS team and getting access to the streams channel – Please be patient 🙂 ).
ü Access the course code repository: Click “Code Repository” on the navigation panel.
ü Check Access to AWS Educate and register for an account: Follow instructions (step 1) at Introduction to Amazon Web Services (AWS) Classrooms
Ø (optional) Install python on your local machine
Ø Refresh your Math and Python skills
COSC2673 | COSC2793 Week 1: Introduction 19
RMIT Classification: Trusted
Course Workload and Assessment
Assessment:
Ø Assignment 1 (30%)
• Individual
• Weeks 3-6
Ø Assignment 2 (50%)
• Group
• Weeks 7-12
Ø Assignment 3 (20%)
• Individual
• Weeks 10-14
Late penalty for assignments:
• Standard CS, -10% each day
• Set last day for submission
• Special Consideration may result in equivalent assessment, not extensions
Workload:
Ø Contact hours: 4.5 hours per week
• Pre-recorded videos (1.5 hrs)
• Lecture Q&A (1 hrs)
• Tutorials (1 hr)
• Laboratories (1 hr)
Ø Self study: 6 hours a week
COSC2673 | COSC2793 Week 1: Introduction 20
RMIT Classification: Trusted
Resources & Reference
COSC2673 | COSC2793 Week 1: Introduction 21
RMIT Classification: Trusted
Text and Reference Books
Primary Reference:
Introduction to Machine Learning, Ethem Alpaydin, MIT Press 2015, 3rd Edition
COSC2673 | COSC2793 Week 1: Introduction 22
RMIT Classification: Trusted
Other References
Additional Textbooks:
Deep Learning, Ian GoodFellow, Yoshua Bengio and Aaron Courville, MIT Press, 2016
(http://www.deeplearningbook.org/)
Pattern Recognition and Machine Learning, Christopher Brown, Springer, 2006
(https://www.springer.com/gp/book/9780387310732) Machine Learning, Tom Mitchell, McGraw Hill, 1997
(http://www.cs.cmu.edu/~tom/mlbook.html) More information on canvas.
COSC2673 | COSC2793 Week 1: Introduction 23
RMIT Classification: Trusted
Tools and Infrastructure
• AWS Educate platform
•
• •
The labs and some assignments are compatible with AWS SageMaker notebooks. (No credit card or payment required)
A classroom for Labs has been already setup and activated in week 1.
More information on setting up and joining an AWS classroom will be available through canvas. See Week 0 Getting started module on canvas.
We will not concentrate on AWS specific features.
The students are encouraged to use this as an opportunity to learn AWS for ML. There are plenty of examples in SageMaker and online.
•
•
COSC2673 | COSC2793 Week 1: Introduction 24
RMIT Classification: Trusted
Communication Channels
Us → You
Announcements are made regularly on Canvas
◦ Please check Canvas regularly Emails will be sent out on a need basis Lecture announcements
You → Us
Consultation hours, email, before/after class, during labs/Tutorials
Official consultation hours: Wednesday’s 4.00pm – 5.00pm (via group MS Teams -> consultation channel)
Many ↔ Many
Discussion Board on Canvas
COSC2673 | COSC2793 Week 1: Introduction 25
RMIT Classification: Trusted
Academic Integrity
Ø Please see:
• https://www.rmit.edu.au/students/student-essentials/rights-and-
responsibilities/academic-integrity
Ø You are encouraged to form groups to solve problems; however, when writing or programming, write in your own words or code, and provide your own solutions.
Ø This is not about penalising or yielding the stick, but you’ve made a conscious act to enroll in this course (thank-you), and we want you to:
• Learn by trying things, making mistakes
• Fair for everyone
• Quality control of your degree
COSC2673 | COSC2793 Week 1: Introduction 26