University of Leicester
Amelia is investigating how machine learning can be used to process the data collected from single photon imaging detectors, a type of quantum imaging instrument. Quantum imaging using photons allows the visualisation of organic and inorganic samples at a far higher resolution than previously possible. At the moment, the technique generates vastly more information than can be processed by current data analysis systems, which limits the performance of the detector technology. Machine learning is becoming an attractive option for handling big data across scientific fields, but usually requires extreme amounts of training data. By developing software that reduces the need for this large initial data input, greater learnings can be achieved in areas where input data is scarce or hard to come by.
Amelia’s work, which will be supervised by academics from the Schools of Computing and Mathematical Sciences and Physics and Astronomy, hopes to challenge the prevailing consensus that more accurate machine learning must be based on larger initial datasets, by demonstrating its success at analysing quantum imaging data without the need for immense datasets upfront.
Commercialising novel detector systems is vital for the success of any new developments in scientific instrumentation, and through this project Amelia will show how combining the detector with machine learning software solutions will create a robust product in the field with numerous applications across industries.
Amelia is a KTP Research Associate at Photek where she has worked since October 2020. She holds a Master’s degree in Physics from the University of Sussex where she has previously worked as a laboratory assistant teacher for their International summer school. She acted as Communications Officer for the University of Sussex’s Q-soc, their resident physics society. Amelia is passionate about young people getting into science, having worked alongside many A-level students and younger whilst an Outreach officer for UoS.
“I am very excited about the impact of this work in unlocking new insights in scientific fields where large datasets are unrealistic or unattainable. This is especially the case in improving the capabilities of space science instrumentation that is so vital for helping us learn more about our universe.”