Reinforcement Learning in Closed-loop data science

Lead Research Organisation: University of Glasgow
Department Name: School of Computing Science

Abstract

This Ph.D. project will explore the use of reinforcement learning and causal modelling in data science applications which involve closed-loop scenarios for either the data acquisition or the application of the final trained controller.
Aims and objectives.
This is part of the EPSRC New Approaches to Data Science Programme, specifically the project Closed-Loop Data Science. This project is about understanding the impact of unmodelled feedback loops on data science practice, and the incorporation of methods from Control Engineering into Data Science. Potential applications will come from computational biology, travel, music recommender systems, smart cities and finance.
Novelty of the research methodology
This project is about understanding the impact of unmodelled feedback loops on data science practice, which has been an important issue often ignored in the past. It is also novel as it includes the incorporation of methods from Control Engineering into Data Science.

Alignment to Research Council's strategies and research areas, and Collaborators
This is part of the EPSRC New Approaches to Data Science Programme, specifically the project Closed-Loop Data Science.
Potential applications will come from computational biology, travel, music recommender systems, smart cities and finance, so collaborators include:
Glasgow Polyomics, JP Morgan, Skyscanner, Moodagent, Aegean airlines.

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/R513222/1 01/10/2018 30/09/2023
2326975 Studentship EP/R513222/1 01/10/2019 31/03/2023 Valentin Charvet