LLM evaluation pipeline development

While LLMs are claimed to be useful for an ever growing range of tasks, systematic evaluation is required to produce evidence of their actual performance. A core component of an automated LLM evaluation is the data used to prompt the model, i.e. the user’s “side” of the transcript. These inputs can be designed in a … Read more

AI companionship evaluation on Roblox

Experiences centred on AI-powered chatbot “characters” – including some offering forms of companionship – are proliferating on the popular gaming platform Roblox. In parallel, the European Commission has designated Roblox as a Very Large Online Platform (VLOP) under the Digital Service Act as the platform surpasses 45 million average monthly users in the EU, and … Read more

Principled uncertainty quantification for electrical generation data [TAKEN]

ELIGIBILITY: This project is for a student taking the MSc in Statistics and Sustainability or the MSC in Statistics and Data Science. This project will investigate methods to predict electricity generation in Ireland. Importantly, we will make predictions with uncertainty intervals that accurately quantify variability in electricity production. We will compare prediction intervals generated by … Read more

Interactive Medieval Manuscripts

A facsimile page from a medieval manuscript can be made interactive: laid flat on a table and wired underneath with conductive tape it can respond to touch. A visitor could press a line and hear the page read aloud in Latin, then in English. Press an illuminated initial and music starts underneath it. The program … Read more

Compositional Data Analysis for Energy Mix Projections

ELIGIBILITY: This project is for a student taking the MSc in Statistics and Sustainability or the MSC in Statistics and Data Science. Compositional data comprise observations that sum to a fixed value and require special attention for statistical modelling and analysis. Energy mix, the proportion of energy use by fuel type, is an example of … Read more

Bayesian Forecasting of Electric Vehicle Adoption

ELIGIBILITY: This project is for a student taking the MSc in Statistics and Sustainability or the MSC in Statistics and Data Science. This project will focus on comparing Bayesian time-series forecasting methods, motivated by the problem of projecting electric vehicle adoption. Depending on interest, this may involve using probabilistic programming through the Stan programming language. … Read more

Extreme Value Theory and the Climate

ELIGIBILITY: This project is for a student taking the MSc in Statistics and Sustainability or the MSC in Statistics and Data Science. Met Éireann monitors weather at 25 stations, with daily historical data available for multiple climate measures including precipitation, temperature, and wind speed. This project will analyze these data using methods from Extreme Value … Read more