Dr Kevin Bleakley
Since October 2010, I am a permanent researcher at INRIA Saclay. My office is located in the maths department of Université Paris-Sud, 22km south of Paris, close to the village of Orsay and slap-bang in the middle of a forest. The forest is pretty in autumn. Whether I get there every day depends on a random variable called the RER B.
I am a member of the INRIA Select team that works on questions in model selection and machine learning. I work at the interface of biology and statistics, and am always looking for new and interesting data-driven mathematical questions that flow from cutting-edge biology research.
From 2008-2010, I was a Postdoc in computational biology with Jean-Philippe Vert at the Ecole des Mines in Paris, a group of talented researchers who are now also integrated into the INSERM U900/Mines/Curie Bioinformatics and Biostatistics Lab at the Institut Curie, a major cancer research centre and hospital in Paris.
Going further back in time, I defended my Ph.D. at Montpellier 2 University, France in November 2007. It involved creating and applying powerful prediction methods to new and exciting biological questions. My supervisor was Gérard Biau. I also collaborated with Marie-Paule Lefranc at the Institute of Human Genetics, among others. Here is a copy of my thesis.
BIOLOGY. I'm using and adapting recent advances in mathematics to create powerful new methods of prediction in various biological situations. I am serious in looking for real solutions to real biological problems which can really help the biologist - either in time savings or money savings. In particular, I am using kernel methods and sparsity-based approaches:
KERNEL METHODS applied to biological problems. These are sophisticated (and very cool) methods designed to extract knowledge from large data sets. These can be particularly powerful in biology because we can encode pertinent information specific to each biological question into the kernel. With these methods we are going to find the cure for cancer and AIDS and save the world from terrorists.
SPARSITY-BASED APPROACHES are ways to perform variable selection and supervised learning in cases where we have too many variables and not enough individuals. In extreme cases such as genomics, this could mean three billion variables and fifty individuals. With a massive dose of prior biological knowledge, we have hope.
New video! It is rumoured that I come from the faraway land of New Zealand where hobbits and elves roam freely amongst the snow-blue lakes and mountains.
As the following conversation from the current movie Wedding Crashers shows, other people have different memories of New Zealand. Bumpy-nosed Owen Wilson is trying to trick the lovely Rachel McAdams into believing he knows how to sail a sailboat.
Rachel : Are you OK?
Owen : Yeah, I'm just swinging the jib for your dad, starboard.
Rachel : But starboard's *this* way.
Owen : That's right. What am I thinking? I'm used to sailing down under with the kiwis so everything's backwards and the toilets when you flush them, the water spins the opposite way. Really freaks you out the first time you see it.
Piha Beach - New Zealand
Getting to a beach like this from Europe is a time consuming exercise. As well as spending 24 hours in a plane (at least!), you also skip a calender day. If this doesn't put you off, try Mike Steel's advice (part 5) for getting through the trip and coming out the other side as a human being. A dose of kiwi English might help too.