Wednesday Seminars

Abstract to be confirmed. Link to join virtually: https://cam-ac-uk.zoom.us/j/81322468305 This talk is being recorded.

Despite immense innovation pressure in the industry, we are held back by the slow evolution of our CPU-focused stand-alone compilation toolchains. Building a new domain-specific compiler, writing a new verification tool, optimizing an application, designing a microprocessor, or verifying some of its components: each of...

The dynamics of weather systems are among the most complex physical phenomena on Earth, and each day, countless decisions depend on accurate weather forecasts, from deciding whether to wear a jacket or to flee a dangerous storm. Until recently, the dominant approach for weather forecasting was “numerical weather prediction...

The machine learning tool box has revolutionized our ability to design novel molecular entities (e.g. proteins) well beyond what the natural repertoire has explored. Despite the incredible advances, the de novo generation of functional molecules in biological concepts remains an incredible challenge. In this talk I will...

Many applications in modern computing (AI, ML and data science) rely on the data from many individuals. However, this data is often sensitive, and needs to be handled under strong guarantees of privacy and security. I'll discuss a range of approaches that can be used at scale to gather statistics and build models of data...

Large Language Models (LLMs) have seen a lot of interest from all over the world, especially since ChatGPT became the fastest growing consumer internet app in history. As we enter a new era of possibilities with AI, new challenges also present themselves. In July of 2023, Meta open-sourced the largest language models to...

Across linguistic theories, human language structures are represented by graphs. Much research has focused on the linearisation of such graphs into actual sequences expressing utterances, but less attention has been paid to the shapes that the graphs themselves take: their topology. A current hypothesis from...

Abstract: Distribution shift is a naturally occurring phenomenon that manifests in various forms of real-world data when the underlying distribution of the data changes over time or between different contexts. This is especially important in the area of human-centered AI where distribution shifts could have important...

We present a new paradigm of designing quantum algorithms via a general-purpose machinery for boosting quantum computation. The framework relies on deep mathematical tools including the quasi-polynomial Bogolyubov-Ruzsa lemma and sheds light on an inherent connection between additive combinatorics and the linearity of...

Voevodsky’s univalent foundations and homotopy type theory comprise a new and radical approach to the foundations of mathematics in which the structural properties of equality are extended to every possible form of equivalence and symmetry, leading to an expanded universe of discourse in which ordinary sets and algebraic...