Schmidt AI in Science

Support for the Schmidt AI in Science Fellowship

 

Schmidt Sciences

The Eric and Wendy Schmidt AI in Science Postdoctoral Fellowship (a programme of Schmidt Sciences) at Oxford is part of a new international initiative to drive innovative use of Artificial Intelligence (AI) in STEM research (engineering, and the natural and mathematical sciences). At Oxford this fellowship includes support from dedicated RSEs within the OxRSE team, which covers

  • Immediate support on accute software problems, e.g. getting software installed and running on ARC
  • Longer term support on ongoing software problems or long-term questions
  • Activate development support for fellows' software projects

Currently supported projects include:

  • Bee Behaviour Tracking: A deep learning system to track bees in experimental chambers in order to analyse how they move and interact with one another. The research aim is to analyse the effects of pesticides on their behaviour.
  • SpeedyWeather.jl: A global atmospheric model with simple physics developed as a research playground with an everything-flexible attitude as long as it is speedy. It is easy to use and easy to extend, making atmospheric modelling an interactive experience.
  • What We Dont Catalog: A project that aims to find unrecorded data features in galaxy catalogues. Work has investigated various methods of visualising the latent space embeddings of autoencoding deep neural networks to be able to understand the feature space.
  • Reef Motion 3D: A project that studies the behaviour of triggerfish in the wild. To analyse the behaviour of these fish from underwater videos, deep learning object detection and tracking algorithms are being evaluated to follow how they move around coral reefs.
  • Morphing Birds: A project to visualise the principal components of bird flight using data collected using motion capture techniques. To disseminate the results of the research to the wider community work has been done to create webpages containing animated graphs where viewers can select individual components in isolation and see how they relate to movement of a model.
  • Meta OpSim: A new project involving Photonics and metasurface optimisation. This is in two strands, one implementing a new field model into an existing metasurface optimiser and the other exploring the possibility of replacing the optimiser entirely with some form of deep learning.

 

Website: https://saiis.web.ox.ac.uk/home

RSEs: Jack Leland, Oliver King