Thu 08 May 2025 - The IFoA’s Federated Learning Working Party will present a novel way for insurance companies to build machine learning and AI models together, without sharing any customer data.
This sessional will walk through the working party’s recently highly commended paper for the 2024 Brian Hey prize. The meeting will show how it has applied Google’s federated learning algorithm (2016) for text prediction on smart phones to insurance modelling.
This concept enables the direct training of machine learning models on users’ devices, such as smartphones. It eliminates the need to share or transfer potentially sensitive data to a centralised server.
Unlike traditional machine learning methodologies, federated learning adopts a model where the algorithm is brought to the data, rather than transferring the data to the algorithm. It will be hugely important as AI becomes more commonplace.
In our paper and sessional we show how insurance companies can collaboratively develop a neural network model to predict claims frequency specifically. We achieve this using the Flower package in Python along with PyTorch. We show that if companies cannot share customer data, they can achieve near double their model predictive performance by using federated learning while still keeping their customer data secure.
The working party is part of the IFoA Data Science and AI practice area.
Dr Małgorzata Śmietanka is an actuary with a PhD in computer science and a researcher at University College London. She chairs the Federated Learning Working Party at the Institute and Faculty of Actuaries and leads the external engagement pillar, fostering collaboration between industry and academia. She specialises in large language models and privacy-preserving AI, with a focus on practical implementations in insurance industry use cases.
Dylan is the Deputy Chair of the Federated Learning Working Party, Explainable AI Working Party, and the IFoA Data Science Research section. His background spans bancassurance, life, and health with a focus on pricing. He currently works at Bupa as the Head of Technical Pricing & Data.
Harry is an experienced actuary and quantitative risk professional with over 12 years of experience in various actuarial roles within consultancies and direct insurers. He has a MSc in AI (Imperial College London, 2020), EMBA (Quantic School of Business 2021), BSc Actuarial Science and Maths (University of Southampton, 2011) and is a Fellow at the IFoA. He also volunteers at various working parties in the IFoA UK and has given talks on AI risk.
Scott Hand is an actuarial analyst with over two years of experience in the field. At Legal & General, he works in the commercial analytics and risk modelling team, where he develops predictive models helping set retail protections pricing assumptions and solve business problems, such as improving the underwriting journey for applicants. Scott holds a Master's degree in statistics from the University of Warwick. Outside of work, he is a keen footballer.
Yung-Yu Chen (Michelle) is a general insurance qualified actuary specialising in capital modelling and currently working in Tokio Marine Kiln. Prior to TMK, Michelle had worked in AIG Shanghai and London offices in the London market.
| Members | Book for free |
| Non-members | Book for free |