Archive for Gamalost

LCM 2012, Trondheim (2)

Posted in pictures, Running, Statistics, Travel, University life with tags , , , , , , , , on June 1, 2012 by xi'an

Today was a much more rewarding day for me at LGM2012 as I was well-rested [and well-fed, thanks to a fantastic breakfast on brunost and gamalost cheeses!]. Indeed, I was able to follow almost all talks without micro-dozing intermissions and I particularly enjoyed those connected with ecology problems, incl. the one on musk oxen. Those morning talks also led me to ponder about the relevance of modelling the corresponding phenomena via Gaussian processes, beyond the computing advantages brought by the Gaussian nature of the underlying process and the availability of INLA. (One could rightly wonder what an agnostic like me is doing at an LGM conference: however, I find those new perspectives on spatial modelling and computing quite refreshing, as well as bringing me in touch with new communities.) Before getting any further into criticisms, I think I should read more carefully about those processes, for instance through Carl Rasmussen and Chris Willliams’  Gaussian Processes for Machine Learning… The afternoon saw more methodological talks with David Dunson speaking about Gaussian processes as a basis for Bayesian non-parametrics (and their frequentist convergence properties), which was quite interesting (although I wish David had given one talk instead of the equivalent to three talks!), and Chris Williams discussing the approximation techniques for Gaussian process regression,  another broad and informative talk! The poster session was also highly diverse, from many case studies to methodological entries on compound likelihood, Bayesian model choice and determinant fast computation. However, I did not stay till the end to go running along the beautiful coast of the Tromdheimsfjord, around the Lade peninsula, enjoying the sun that had showed up in the late afternoon if not the fierce and cold wind…

Incidentally, I learned today (during Dan Simpson’s reinvigorating talk) that Cholesky was a French mathematician (who died at the very end of WWI, at the age of 43, as a colonel in the French artillery, and whose famous decomposition got published posthumously by a fellow officer).