Introductory description

This module introduces students to the contemporary practice of Machine Learning and Deep Learning. The module takes a hands-on approach where concepts are introduced using the various Machine Learning and Deep Learning frameworks developed in Python. This module is offered as an optional module to Statistics students, and as an unusual option to students from other departments, space permitting.

Prerequisites. This module assumes you have studied and completed ST231 Linear Statistical Modelling with R,

Recommended. It is recommended that you have studied and completed ST246 Python for Data Analytics Tasks and ST340 Programming for Data Science.

Module aims

Outline syllabus

This is an indicative module outline only to give an indication of the sort of topics that may be covered. Actual sessions held may differ.

This module covers the following topics.

1. Modes of statistical learning / Common Machine learning tasks. Unsupervised Learning; Semi-supervised learning.

2. Machine learning frameworks. Machine learning frameworks in Python (Scikit-Learn, JAX, graphlearning).

3. Unsupervised learning / dimensionality reduction algorithms: for example, principal component analysis, singular value decomposition, t-SNE embeddings, k-means clustering, spectral clustering, Fokker-Planck clustering, Hierarchical clustering, mixture models, DBSCAN, OPTICS.

4. Semi-Supervised Learning algorithms: for example, Transductive Support Vector Machines, Modularity MBO, Generative Models, Nearest Neighbour, Laplace Learning, Poisson Learning, Poisson Learning, Centered Kernel Method, p-Eikonal Classifier

Learning outcomes

By the end of the module, students should be able to:

• Appraise and decide on the appropriate type of statistical learning required to address a real-world challenge.
• Create data-analytic pipelines and workflows for the various modes of statistical learning.
• Implement common Machine Learning and Deep Learning frameworks in Python.
• Collaborate and disseminate fully documented Python code with reproducible outputs.
• Interpret and communicate the outputs of a completed analysis to a range of audiences.

Indicative reading list

Reading lists can be found in Talis

Interdisciplinary

This module requires students to develop a balanced facility of familiarity with machine learning frameworks and data-analytic skills for solving real-world problems across disciplines.

Subject specific skills

1. Demonstrate advanced facility with data handling and analysis methods in R and Python.

2. Create readable, valid, reliable, reproducible and well-documented code.

3. Appraise problems, abstracting their essential information to make judgements on the appropriate concepts to facilitate their solution.

4. Demonstrate programming skills and knowledge of programming concepts, both explicitly and by
   applying them to the solution of real-world problems.

Transferable skills

1. Problem-solving skills: The module requires students to solve problems and present their conclusions as logical and coherent arguments.

2. Written communication skills: Students complete written assessments that require precise and unambiguous communication in the manner and style expected in mathematical sciences.

3. Verbal communication skills: Students are encouraged to discuss and debate formative assessment and lecture material within small-group tutorial sessions. Students can continually discuss specific aspects of the module with the module leader. This is facilitated by statistics staff office hours.

4. Team working and working effectively with others: Students are encouraged to discuss and debate formative assessment and lecture material within small-group tutorial sessions.

5. Professionalism: Students work autonomously by developing and sustaining effective approaches to learning, including time management, organisation, flexibility, creativity, collaboratively and intellectual integrity.