ENGG3300Machine Learning for Engineers

Course handbook

Description

This introductory course on statistical machine learning is intended to offer foundational concepts in both the theory and practice of this exciting area. It is intended for students with some background in physics, calculus and linear algebra and focuses on supervised learning, i.e., classification and regression, and introduces nonparametric Bayesian methods. A major theme of this course is to introduce flexible model structures, and incorporate fundamental limits (such as physical conservation laws) - where appropriate - so that predictions obey these limits. The course will cover a range of methods used in statistical machine learning and will review relevant areas of statistics and probability theory as required. These methods will be studied and applied to real data from various applications throughout the course. The course also covers important practical considerations such as cross-validation, model selection and the bias-variance trade-off. The course includes theory (e.g., derivations and proofs) as well as practice (labs and assignments) and is delivered using a problem-based format. The practical aspects are implemented using the industry standard programming language, Python, which will be introduced in context throughout the course.

Availability2024 Course Timetables

Learning outcomes

On successful completion of the course students will be able to:

1. Propose appropriate flexible models for different problem domains

2. Quantify the uncertainty of predictions and estimates

3. Implement solutions in Python

4. Select a model using cross-validation and bias-variance trade-offs

5. Design and train deep neural-networks for classification

6. Construct Gaussian Process models for nonparametric modelling

7. Apply machine learning methods to large-data problems

8. Design solutions that adhere to fundamental limits, where appropriate

Content

1. Review of relevant topics in statistics and probability
2. Linear regression (including ridge regression and the Lasso)
3. Classification via logistic regression and k-nearest neighbours
4. Linear and quadratic discriminant analysis
5. Regression & classification trees (including bagging and random forests)
6. Boosting
7. Neural networks and deep learning
8. Bayesian non-parametric methods (including the Gaussian process)
9. Incorporating fundamental limits into solutions
10. Introduction to Python

Assumed knowledge

MATH1110 Mathematics for Engineering, Science and Technology 1, MATH1120 Mathematics for Engineering, Science and Technology 2, MATH2310 Calculus of Science and Engineering, ENGG1003 Introduction to Procedural Programming.

Assessment items

Tutorial / Laboratory Exercises: Laboratory Exercises (x10)

Written Assignment: Assignments (x2)

Contact hours

Trimester 3 - 2024 (Singapore) - Singapore PSB

Semester 2 - 2024 - Callaghan

Course outline

Course outline not yet available.