About this event 

There has been a revolution in Bayesian inference, the new Hamiltonian / Hybrid Monte Carlo samplers have made it much easier to apply Bayesian approaches. This is great because they work well for small data sets. Cell count data in neuroscience is difficult to gather, sometimes data from as few as six animals is available so they have a tiny "n-number"; however, there is a lot of data for each animal, a count for each brain region. In this talk I will describe how a partial pooling Bayesian model can find results in these data in cases where more traditional statistical approaches fail. Although this is a neuroscience example, the approach will be useful for anyone with "rectangular" data, that is, data with lots of variables but very few measurements.

Please look at our Code of Conduct which we follow at all virtual and in-person events.

Audience Level:

Intermediate

Learning Outcomes:

An understanding of what Bayesian inference can bring to small data sets.

Speaker

Dr Conor Houghton, Associate Professor of Computer Science, School of Engineering Mathematics and Technology

Dr Conor (Con) Houghton (he/him)

I am a computational cognitive scientist with a particular interest in linguistics, neurolinguistics and artificial intelligence. I started my career in mathematical particle physics, I did a PhD in Cambridge on three-dimensional field theories with soliton solutions and stayed on there as a research fellow before doing a postdoc in Columbia New York. In 2000 I began as a lecturer in theoretical physics in Trinity College Dublin (TCD), I was glad to be back in Ireland since I grew up near Galway in the west of Ireland, I also loved TCD; however, I became interested in neuroscience and ultimately, in 2012, I moved to Bristol because it has a strong neuroscience community and a history of collaboration between experimental and computational neuroscientists.

Bristol Data Week 2024

This event is part of Bristol Data Week 2024, organised by the Jean Golding Institute. Running from Monday 3rd June – Friday 7th June 2024, this will be our 7th annual Data Week; an interactive programme of speakers, training and workshops open to all and completely free of charge.

Keep up to date with sessions happening throughout Bristol Data Week on the Jean Golding Institute website follow us on Twitter @JGIBristol and use #BristolDataWeek.