July 2019 - Digitizing Biology

This topic occurred to me following my recent talk at a dental conference at Charité Berlin. Upon hearing that I have a strong interest in inference, my fellow keynote mentioned that it drives him crazy that random forests, and similar algorithms, work so much better than DNNs on genomic data. He challenged me to come up with a reason for why this is the case.

I think that I know why. The problem I have is that I suspect that I can never prove it. That issue of not being able to prove things in machine learning is probably an equally interesting topic, for a future article, but here I want to address my theory of why random forests work better than DNNs for analysing genome data.

A sense of home is a powerful feeling. The sense of belonging, of knowing where everything is. I miss that sometimes.

I left Ireland almost exactly 10 years ago with a burning need to go out and prove myself. I had finally recovered from Chronic Fatigue Syndrome and was going to take up a much delayed PhD position. I moved to the University of Luebeck, where I found my introduction to neuroscience, before moving on to Paris Descartes, the École Normale Supérieure and the University of Chicago. With each move I developed a new network of colleagues, collaborators, and mentors.

A couple of weeks ago I was contacted by a member of the Irish mathematical community. Many years ago I wrote the website for the Mathematics Department at NUI, Galway. While doing that I included biographies of the then members of staff and their research domains. At some point, I also made a backup of the website for my own reference under my personal domain. Sadly, many of the members of staff who worked at NUI, Galway when I was an undergraduate are now dead. So now this resource has become a useful archive. And thus I was re-discovered by a member of the current Irish mathematical community.

Month: July 2019

Why do Trees work better than DNNs on genome data?

The network you grow up with