Causality and Compositionality

Events, Seminars, Conferences and more at the School of Engineering.

School of Engineering

Join us for a deep dive into the interconnected worlds of causality and compositionality online!

Speaker : Professor Ricardo Silva,Professor Ricardo Silva earned his PhD in Machine Learning from Carnegie Mellon University as part of the program’s inaugural cohort. He then moved to the UK for postdoctoral research at the Gatsby Computational Neuroscience Unit at UCL, followed by a year at the Statslab in Cambridge. In 2008, he joined the Department of Statistical Science at UCL as a Lecturer. A leading expert in causal inference and AI, Professor Silva currently holds an EPSRC Open Fellowship. He is Co-Lead at CHAI Hub, particularly as a Research Theme Lead, and collaborates with industry partners, including DeepMind, Spotify, and ByteDance. <h2> Causality and Compositionality</h2>Compositionality has become some kind of buzzword in AI. In a broad sense, compositionality is the ability to respond to combinations of situations that have not appeared together before, particularly if in each isolated situation an agent has had the experience of what to do. In a more down-to-earth sense, it essentially boils down to tasks of combinatorial extrapolation.In the context of core causal inference, we can think of the task of predicting the effects of combinatorial interventions, each of which can be understood as a switch: from a “off” regime (observational) to some other “on” regime where a system has been poked by a variety of physical means. For instance, a cell can be regulated by perturbing different sets of components; the user of some service can be shown different combinations of options, which we hope will influence their behaviour; and so on.There is no single way of achieving this effect compositionality. I will describe a sample of three: i) exploiting additive confounding structure that allows the predictions on combinations of interventions from interventions in single variables; ii) pushing the idea of causal graphs into a more abstract direction via structured energy-based models; iii) exploiting structured function spaces that map actions to their effects, including in a longitudinal setup

Our {communityGuidelinesLink} describe the sort of content we prohibit on Eventbrite. If you suspect an event may be in violation, you can report it to us so we can investigate.

To report other categories of prohibited or illegal content, submit {link}

Causality and Compositionality

More events from School of Engineering

Discover more events from School of Engineering, from Science & Tech to other experiences you might love.

Still looking for the right event?

Explore all online events to browse and filter by date, category, and more.

Causality and Compositionality

More events from School of Engineering

Discover more events from School of Engineering, from Science & Tech to other experiences you might love.

You might also like...

Browse more events with different dates, prices, and formats to find your next great experience.

Still looking for the right event?

Explore all online events to browse and filter by date, category, and more.