Scholarship makes it possible to present their functional inverse diffusion algorithm abroad
KTH Opportunities Fund
Pol del Aguila Pla from the Department of Information Science and Engineering has received a scholarship from KTH Opportunities Fund. The money will be used to present their functional inverse diffusion algorithm to the research community in the US, South Korea and the UK.
Congratulations! Let's start with the classic question: how does it feel?
- Thanks, it's fantastic! I think it must be one of the best feelings in the world, writing a project, arguing why it is so important, and then having someone actually believing in you and giving you the resources to do it. It is really great, but a bit scary too.
Tell us about the work you've received the scholarship for. What is it about?
- So, 2 years ago, and thanks to a former Ph.D. student at the department, Klas Magnusson, a Swedish biotech (Mabtech AB) contacts us to do some research. They wanted us to figure out how to analyse the data in some biochemical assays, Elispot and Fluorospot, which they sell material for. These assays are used for tons of things in bio-sciences: developing vaccines, diagnosing diseases, etc. After a lot of thinking, we came to see it as an inverse problem in function spaces, so we developed what we call a functional inverse diffusion algorithm. Long story short, we have now solved their problem much better than they expected by running so-called matrix-free optimisation algorithms on modern hardware for graphics processing (GPUs). These are algorithms that can handle very large data structures and problems, in our case, 34 million variables!
What possibilities come with receiving this scholarship?
- Now, that the company has patented our findings and we can finally share them, we will first publish everything. Then, we will use the resources for presenting our work to the different scientific communities where it is of relevance. One of the best things of what we got is that it goes beyond its importance to the biosciences, and also contributes to research in other fields like signal processing and applied mathematics. The plan now is to present it to the community that develops methods for bioimaging (ISBI 2018, Washington DC, USA), to the signal processing community at ICASSP 2018, Seoul, South Korea and to the applied mathematics community at the workshop "Generative models, parameter learning and sparsity" at the Isaac Newton Institute for Mathematical Sciences in Cambridge, UK. Wish me luck!