Artificial Intelligence for Science MSc

Computational methods using artificial intelligence (AI) and machine learning (ML) offer powerful applications across the diverse landscape of science. This module offers an introduction to the fundamental principles and practical applications that underpin AI and ML, specifically tailored to address challenges and unlock discoveries within the biosciences, mathematical and physical sciences, and psychology. The module covers core AI concepts and the essential mathematical foundations of ML. Hands-on exercises equip students with the skills to implement AI and ML applications to scientific datasets using relevant programming languages. A significant component of this module involves a collaborative, intensive project where students work in teams to tackle a scientific data challenge and generate meaningful insights and potential solutions. By the end of this module, students will be well-equipped to leverage the power of AI and ML to address complex scientific questions and contribute to data-driven discovery in their chosen field.

This module explores the current frontiers of AI as applied to scientific questions, and includes a critical evaluation of the issues raised. Students will participate in workshops and seminars led by researchers across the life and physical sciences, examining how AI methods are transforming their research and shaping employment opportunities. Topics covered may include, for example: predicting protein structures from amino acid sequences; automatic detection and monitoring of biodiversity from images or acoustic data; analysis of behavioural data; natural language processing for text analysis; neuroimaging analysis; image analysis, time series analysis and classification tasks in astronomy; optimisation of large (or expensive) models in theoretical chemistry; connections between ML methods and statistics. Critiquing the technical and ethical issues raised by AI applications will also form an important part of this module, which will also cover the impact of emerging issues such as data bias, model interpretability, reproducibility, and environmental impact, as well as how researchers are using AI for more generic tasks such as systematic searching of the literature and use as a coding assistant.

This module provides basic skills in computational data analysis. Students will learn how to import/export scientific data sets in different formats, how to process and transform them, and how to visualise results. Teaching will be hands-on and computer lab-based. Teaching will focus on the programming language R and associated scientific software. No prior programming experience will be necessary.

This module will investigate a range of ethical issues pertaining to current and ongoing developments in AI and contemporary technology. Example topics might include responsibility dilemmas in regard to autonomous vehicles and lethal autonomous weapon systems, relationships with chatbots and robots, data ethics and algorithmic injustice, as well as the moral status of future artificial beings. By studying these topics and others, students will gain a broad philosophical perspective on advanced topics in AI and technology ethics, develop their ability to critically assess and face moral dilemmas in the dynamic and fast-paced environment of the AI and tech scene, as well as improve their understanding of how philosophical theories can be applied and used to tackle practical real-world ethical challenges posed by modern and future technology.

This module gives students the opportunity to develop to a high level skills relevant to a career applying AI methods to address scientific problems. Students will conduct a desk-based computational project, with potential to as part of a small team addressing a larger scientific problem, and/or to work with an external organisation. Common elements include the independent planning and production of an original piece of research, under the guidance of an academic supervisor. The project write-up is targeted to a specific audience - either a scientific research or review paper, or a report aimed at a specific sector.

This module considers the analysis of data in which the same quantity is observed repeatedly over time (e.g., recordings of the daily maximum temperature in a particular city, measured over months or years). Analysis of such data typically requires specialised methods, which account for the fact that successive observations are likely to be related. Various statistical models for analysing such data will be presented, as well as how to implement them using the programming language R.

This module starts with a primer on neuroscience and the role of computational neuroscience. The module will cover various modelling approaches, from classic biologically plausible to abstract-level models of neurons. The module will then move to higher levels of modelling approaches, such as neural networks and reinforcement learning. While the module emphasises methodological issues and how models can be built, tested and validated at each level, we will also draw connections to specific brain regions to motivate and illustrate the models.

This module explores cutting-edge applications of artificial intelligence that are revolutionising Biosciences. Students will gain hands-on experience in using AI tools and techniques to analyse biological data, model complex systems, and accelerate scientific discovery. Through practical examples and case studies, the module will show how AI is transforming various aspects of Biosciences. Case studies may include applications in areas such as cell and molecular biology; physiology, development and disease; and biodiversity science. Cross-cutting applications including machine learning, computer vision, and natural language processing in biosciences may also be covered. Specific applications in topics such as predicting biological structures and automatically classifying images or audio recordings will give students hands-on experience and enable them to develop skills that are increasingly required by employers in the biosciences. Students will end the module by completing and writing up a practical challenge.

This module develops the Bayesian approach to statistical inference. The Bayesian method is fundamentally different to the approach taken in earlier statistics courses. It is a more general and more powerful approach, and it is widely used, but it relies on modern computers for much of its implementation. It is based on the idea that if we take a (random) statistical model, and condition this model on the event that it generated the data that we actually observed, then we will obtain a better model. This course covers the foundations of Bayesian statistics and the incorporation of prior beliefs, as well as computational tools for practical inference problems, specifically Markov Chain Monte Carlo and Gibbs sampling. Computational methods will be implemented using R and Python. Advanced computational techniques will be explored, in the second semester, using STAN.

This module covers advanced statistical techniques increasingly required in psychological research, specifically confirmatory factor analysis, structural equation modelling, multilevel modelling for both cross-sectional and longitudinal data, and generalised linear models. Lectures will be used to teach the rationale and principles behind these techniques, with practical sessions offering the opportunity to apply and develop students' knowledge. The course will use the statistical environment R.

This module concerns inferring and modelling neural and cognitive processes underlying human behaviour using computational means. One part of the module will cover normative models, which allow us to solve problems optimally along with their neural or cognitive representations. The other part of the module will focus on cognitive models, which involve fitting models to behavioural data to estimate latent parameters that are assumed to underlie the data and allow us to make inferences about their properties.