Seminars

PhysLean is a project to create a monolithic library of results from physics in Lean 4, akin to Mathlib for mathematics. It contains results from a range of different areas of physics including classical mechanics, relativity, condensed matter physics, quantum field theory, string theory etc.

Despite massive investments in training large language models, tokenizers remain a critical but often neglected component with weaknesses that can cause wild hallucinations, bypass safety guardrails, and break downstream applications. This talk will cover:

The Constraint Satisfaction Problem (CSP) is a type of decision problem with several equivalent formulations. Its original definition was inspired by considerations in Descriptive Complexity, and represents a large part (in some sense) of the class NP.

Federated Learning (FL) has emerged as a key paradigm for enabling collaborative and privacy-preserving machine learning across distributed data sources.

Dashboards are a critical part of your observability stack. Looking at your dashboards—are they actually helping you understand the system? When well-designed, they surface intuitive information and help quickly diagnose outages.

For more details about the talk, and if you would like to attend, please visit: https://www.eventbrite.co.uk/e/talk-by-professor-bjarne-stroustrup-concept-based-generic-programming-tickets-1742215837469?aff=oddtdtcreator

Companies need reliable emission data for their products and services to take effective climate action, yet obtaining it is challenging. In today's interconnected economy, product and service carbon footprints (PCF) cannot be determined in isolation but require emission data exchange throughout supply chains.

Abstract not available

By 2050, older adults will make up about 22% of the global population, driving an urgent need for accessible and reliable health technologies. In this talk, I will present our work on intelligent mobile systems designed for older adults.

Federated Learning (FL) is a paradigm where models are collaboratively trained by sharing only local parameters with a central aggregation server and faces limitations in heterogeneous environments.