ML Foundations Course

An introduction to machine learning foundations, designed as a 4-week course with 3 sessions per week. The course will be first run starting 1st September 2025.

Course Structure

Duration: 4 weeks
Sessions: 3 sessions per week
Start Date: 1st September 2025

Topics and Resources

Week 1: Foundations

• Session 1: Probability and an Introduction to Jupyter, Python and Pandas
• Session 2: Objective Functions: A Simple Example with Matrix Factorisation
• Session 3: Linear Algebra and Linear Regression

Week 2: Basis Functions and Generalisation

• Session 1: Linear Algebra (continued)
• Session 2: Basis Functions

Week 3: Neural Networks

• Session 1: Generalisation and Neural Networks

Week 4: Advanced Topics

• Session 1: Bayesian Regression
• Session 2: Gaussian Processes

Week 5: Applications and Systems

• Session 1: Latent Variable Modelling

Core Topics Covered

• Probability and probability distributions
• Linear algebra and linear regression with appropriate factorisations
• Basis functions and feature engineering
• Generalisation and double descent
• Bayesian regression
• Gaussian processes
• Latent variable modelling and representation learning

Lecture Materials

All lecture materials are available in multiple formats:

• Web Lectures: View lectures online
• Slides: Download presentation slides
• Jupyter Notebooks: Interactive notebooks
• Practical Exercises: Hands-on exercises

Available Lectures

1. 01-01-probability: Probability and an Introduction to Jupyter, Python and Pandas
2. 01-02-matrix-factorization: Objective Functions: A Simple Example with Matrix Factorisation
3. 02-01-linear-algebra: Linear Algebra and Linear Regression
4. 02-02-basis-functions: Basis Functions
5. 03-01-generalisation-and-neural-networks: Generalisation and Neural Networks
6. 04-01-bayesian-regression: Bayesian Regression
7. 04-02-gaussian-processes: Gaussian Processes
8. 05-01-latent-variable-modelling: Latent Variable Modelling

Datasets

• River water level (time series)
• Weather (TAHMO, time series) - TAHMO
• DSAIL porini (images of wild animals) - DSAIL Porini
• Kieni forest (terrestrial and aerial images)
• Sensor placement problem
• UK census data
• Kenya census data - Kenya Census 2019

Dissemination Ideas

• Video recordings of lectures
• Dedicated web pages with lecture materials on the DSAIL website
• Paper on teaching methods and outcome

Course Delivery

The course is designed to be delivered in a hybrid format:

• In-person lectures with interactive demonstrations
• Online materials for self-paced learning
• Practical sessions with hands-on coding exercises
• Assessment through practical assignments and final project

Prerequisites

• Basic programming experience (Python preferred)
• Familiarity with high school mathematics (calculus, linear algebra)
• No prior machine learning experience required

Team

• Cedric Kiplimo
• Austin Kaburia
• Fred Lawrence
• Neil Lawrence @lawrennd
• Ciira Maina
• Radzim Sendyka

Repository Structure

mlfc/
├── _config.yml           # Jekyll configuration
├── _lamd/                # LaMD files containing lecture sources
│   └── _lamd.yml         # LaMD configuration (author, output formats, paths)
├── _lectures/            # Compiled Lecture html files
├── _notebooks/           # Compiled Jupyter notebooks
├── _practicals/          # Compiled Practical exercises
├── assets/               # Static assets (images, js, css)
├── index.md              # Home page
├── lectures.html         # Lectures index page
├── slides/               # Compiled reveal.js presentation slides
├── backlog/              # Project backlog and task tracking
├── cip/                  # Code Improvement Plans
└── docs/                 # Additional documentation

Usage

Lectures are compiled using the maketalk command:

cd _lamd
maketalk probability.md 

Configuration

The _lamd/_lamd.yml file controls how your lectures are processed and where outputs are stored. Key settings include:

• author: Your personal information
• postsdir, slidesdir, etc.: Output directories for different formats
• posts, docx, pptx, etc.: Enable/disable different output formats
• baseurl, url: Website configuration for GitHub Pages

See the repositories at mlphysical or advds for examples of complete lecture courses using the LaMD system.

Learn More

For more information about LaMD, see the LaMD documentation.