Physics Chat
Machine Learning in Microscope Imaging
Speaker: Matyo Ivanov (JP)
Date: Friday 20 November 2020
Time: 15:00
Venue: Zoom
We investigate the ability of Machine Learning algorithms to aid and improve experimental results by providing otherwise unattainable information to the experimentalist and by making fine adjustments to the experimental setup in real-time. By training a Convolutional Autoencoder on a dataset of Convergent Beam Low Energy Electron Diffraction (CBLEED) images, we show the ability of the algorithm to process abstract experimental data into an interpretable cluster space. Combining this with a Reinforcement Learning algorithm which trains on the Low-Energy Electron Microscope (LEEM) and learns how to optimize it, we aim to reach the otherwise very difficult to achieve optimal imaging conditions required to perform CBLEED imaging.
Join Zoom Meeting https://cardiff.zoom.us/j/82513817850?pwd=NU5HUkpHN1d3U2VoVmRaWDF4M1hMZz09 Meeting ID: 825 1381 7850 Password: 322173
----- Upcoming seminars and colloquia are listed on the internal webpage: http://www.astro.cardiff.ac.uk/newsandevents/?page=seminars&series=physicschat
-- Recordings of previous seminars and colloquia can be found in the 'Physics Chat' folder on Panopto: https://cardiff.cloud.panopto.eu/Panopto/Pages/Sessions/List.aspx#folderID=%22f5986f52-3e46-4910-ae87-ac4e009ff1a7%22&folderSets=1