Foundation Year
The foundation year involves a series of modules covering the five core foundations:
- Data-Driven AI
- Knowledge-Driven AI
- Human-AI Interaction
- Responsible AI
- Interactive AI in Context
You will complete a range of individual and group projects in preparation for your research project. The material is divided into three main teaching blocks - Teaching Block 1 (TB1, Autumn-Winter), Teaching Block 2 (TB2, Spring) and Teaching Block 3 (TB3, Summer).
Machine Learning Paradigms
This unit gives an in-depth overview of Machine Learning, exploring both unity and diversity among different ML paradigms and why this diversity is needed and how it can be exploited. The paradigms covered include: Introduction: tasks, models and features; Tree and Rule models; Linear and Distance-based models; Probabilistic models; Model ensembles; Deep learning. The unit will provide students with a solid analytical and practical framework for further work in data-driven AI.
Computational Logic for Artificial Intelligence
This unit provides an introduction to knowledge-driven AI from the perspective of computational logic. It covers the basic principles of knowledge representation and automated inference by means of logic programming languages, which have pattern matching and backtracking search as primitive operations.
This then leads to more advanced methods in natural language processing and machine learning which exploit the representation and reasoning power of logic programming.
Dialogue and Narrative
This unit presents the theoretical foundations for interactions that are narratively structured, reflecting the human bias towards organising our experiences as (language, logical and visual) narratives. Understanding the narrative structure requires a deep understanding of Natural Language Processing (NLP) components going from low-level text mining and segmentation to discourse-level analysis and summarisation. The unit builds upon these to teach core concepts in dialogue management, argumentation theory, and computational modelling of narratives.
Responsible AI
This unit gives a solid grounding in fairness, accountability, transparency, privacy and trustworthiness in AI, and related concepts relating to ethics, law and regulation. Using case studies we will present and analyse these concepts from the perspective of industry, academia and government. Wherever possible these case studies will be drawn from PhD projects from earlier-cohort CDT students or other PhD students in the school.
Research Methods in Interactive Artificial Intelligence
This unit introduces CDT students to a range of research methods in interactive AI, and prepares them to carry out their own research in the Summer Project and subsequent PhD research. Through the unit students will improve the breadth and depth of their general AI knowledge, learn how to process and present scientific material, and how to plan for a larger research project.
Research seminars given by external and internal speakers will serve as exemplars of state-of-the-art AI research. Students will also research selected topics from the literature and present their findings in oral and written form (first deliverable). The topics will be chosen to be practically applicable and make students reflect about future research directions and about their own research. The second deliverable of the unit will be a Project Synopsis that prepares the ground for the Summer Project.
Applied Data Science
This unit introduces key data science concepts and their application to support data-driven approaches to problem solving. The aim of this unit is to allow students to acquire fundamental skills covering the full data science pipeline, including the pre-processing, manipulation, integration, storage, exploration, visualisation and privacy.
Students will study techniques to transform raw data into advanced representations that will enable a deeper understanding of the original data:
- Data ingress and pre-processing
- Data storage and data management
- Data transformation and integration
- Data exploration and visualisation
- Data sharing, privacy and anonymisation
The students will also gain practical skills in handling structured and unstructured data, gaining hands-on experience of software tools widely used in real-world settings.
Uncertainty Modelling for Intelligent Systems
This unit will explore the techniques and methodologies developed within Artificial Intelligence to represent and reason with information which is uncertain, imprecise or fuzzy. The unit will provide an overview of a range of different approaches explaining both the mathematics and the underlying philosophy and investigating practical applications.
Aims:
- To provide students with an overview of uncertainty modelling techniques and formalisms
- To provide students with an in-depth study of the mathematics and philosophy underlying these techniques
- To provide a detailed analysis of the application of uncertainty modelling in intelligent systems.
Interactive AI Team Project
The aim of this unit is to provide students with practical experience in applying user-centred, theory-informed and evidence-based design and implementation methodologies to a real-world interactive AI challenge. The work will be done in teams of 3-5 students.
Unit content:
- Software engineering methods and tools for interactive AI pipelines
- Approaches to participatory and user-centred design
- Development of low and medium fidelity prototypes
- Approaches to evaluating interactive AI applications
Interactive AI Summer Project
The project offers two possible itineraries. For students with a clear project/supervisory team in mind, this sets the scene for the PhD thesis to be developed in Years 2 to 4. The main purpose of this initial phase is to compile the literature review and analyse the feasibility, social impact and any ethical issues. It will deliver a small proof-of-principle implementation and also a report, including the outline plan of the subsequent project. The project will be presented in the form of a poster in the next Summer School. For students that need to further explore the field, they will be allowed to undertake two smaller projects involving up to two supervisory teams, such that one of them will develop into the subsequent thesis.This flexibility in how to shape the Summer project was strongly recommended by our project partners.
Between TB1 and TB2, a Winter School will be organised involving all CDT cohorts, industry partners and potential supervisors. The aim of the Winter School is to showcase the range of research being done in the CDT, and to help prepare Y1 students to choose their Summer project and PhD topics.