The New Years special edition of Tech Socials presents

Using AI to predict zebrafish brain state and behaviour from whole-brain cellular-resolution functional imaging

Abstract

Behavioural responses vary across presentations of identical sensory stimuli, in part owing to long-term changes in internal state variables such as motivation and stress. Some of these state variables are represented by ongoing neuronal activity, persisting in the absence of sensory input and often involving complex patterns across multiple brain areas. I study how ongoing activity in the larval zebrafish brain is associated with the execution of innate visually-evoked behaviour. I simultaneously track the activity of 100,000 neurons using light-sheet microscopy, a method that enables high-speed volumetric imaging of the entire zebrafish brain in vivo. To identify patterns associated with specific behavioural features, I train artificial recurrent neural networks to predict behavioural outcomes from activity recorded prior to prey-like visual stimulus presentations.

I will discuss some of the technical challenges I'm facing, related to elaborate online control, dealing with immense data volumes, and complicated model interpretation.

Speaker

Asaph is a Postdoctoral Researcher in Dr Isaac Bianco's lab, based in the Department of Neuroscience, Physiology and Pharmacology at UCL. He is studying how internal state and network dynamics shape behaviour, by combining in vivo experimental approaches in larval zebrafish and computational modelling.

Asaph studied Biology for his BSc at the Hebrew University of Jerusalem, and stayed there for his MSc and PhD in neuroscience.

Joining

This event will be held in-person in at ground floor of Bidborough House in UCL. We will follow up with this great talk with Q&A over food and drinks that will bring a holiday spirit! If you'd like to join us remotely, please use the information below:

Microsoft Teams meeting

https://teams.microsoft.com/l/meetup-join/19%3ameeting_OTkwZTY5MmYtNWU4Yi00M2Q5LWJlZmEtNTRkNzhkMWNjNTEy%40thread.v2/0?context=%7b%22Tid%22%3a%221faf88fe-a998-4c5b-93c9-210a11d9a5c2%22%2c%22Oid%22%3a%22b11675b2-b99a-4c04-8023-e40534766a69%22%7d

Meeting ID: 360 569 959 966

Passcode: KWwNNa

Download Teams | Join on the web

Or call in (audio only)

+44 20 3794 0272,,966207753# United Kingdom, London

0800 047 8623,,966207753# United Kingdom (Toll-free)

Phone Conference ID: 966 207 753#

Find a local number | Reset PIN

For more information about the Tech Socials, and an archive of previous events, see our page!

To keep up with announcements about our events, sign up to our mailing list or follow us on Twitter.

Help with joining remotely

This event is hosted in a Microsoft Teams meeting by UCL (see info above). Please ensure that you have the Microsoft Teams App installed. To find out more:

UCL participants: You can access the UCL Teams Support Centre. Sign in to the Teams App using your UCL credentials.

External participants: If you already use Teams, sign in to the Teams App using the credentials issued by your host university, business or other provider. If you do not have the Teams App installed, or are not currently using Teams, please click here to install the Teams App.

You do NOT need to possess an Office 365 or 'Teams Licence' to attend this event. If you do not currently use Teams, you do not need to attempt to 'Sign In' to the App. All external participants should click the meeting link to join the meeting at the time of the event, and choose to 'open it in the App' (if installed).

If you do not want to or cannot use the Teams App, you can try joining through your browser, although please note that only Google Chrome and Microsoft Edge are officially supported.