Breaking down the jargon: key terms in Project Athena
Data analytics and machine learning are central to Project Athena, enabling the programme to identify potential fraud early and more accurately by analysing vast amounts of data. Here’s a quick guide to some essential terms related to the project.
Machine learning
Machine learning is a type of data analysis where algorithms learn from data and improve over time. Project Athena uses machine learning to recognise patterns in fraud data and identify emerging behaviours.
Deep learning
Deep learning is an advanced form of machine learning that uses neural networks to analyse complex tasks, such as recognising intricate patterns. It helps Project Athena detect complex fraud patterns in the data.
Supervised and unsupervised learning
In machine learning, two main approaches help computers learn from data: supervised learning and unsupervised learning. These are key to the work we are doing on Project Athena. Think of them like two different teaching methods.
- Supervised learning: The system learns from labelled examples, much like teaching a child to recognise good and bad behaviour by providing examples of both. In Project Athena, we use past fraud cases to help the model "learn" what fraud looks like, improving its ability to spot similar patterns in new data.
- Unsupervised learning: This approach is more exploratory, like letting a child observe different behaviours on their own and identifying patterns without guidance. In Project Athena, unsupervised learning means grouping data in ways that reveal suspicious patterns, even when we don’t have specific examples to label as fraudulent.
Visit our new glossary to find out about more key terms around Project Athena.
If you have any questions about our work, contact ProjectAthena@nhscfa.gov.uk