brief stop in Edinburgh

Yesterday, I was all too briefly in Edinburgh for a few hours, to give a seminar in the School of Mathematics, on the random forests approach to ABC model choice (that was earlier rejected). (The slides are almost surely identical to those used at the NIPS workshop.) One interesting question at the end of the talk was on the potential bias in the posterior predictive expected loss, bias against some model from the collection of models being evaluated for selection. In the sense that the array of summaries used by the random forest could fail to capture features of a particular model and hence discriminate against it. While this is correct, there is no fundamental difference with implementing a posterior probability based on the same summaries. And the posterior predictive expected loss offers the advantage of testing, that is, for representative simulations from each model, of returning the corresponding model prediction error to highlight poor performances on some models. A further discussion over tea led me to ponder whether or not we could expand the use of random forests to Bayesian quantile regression. However, this would imply a monotonicity structure on a collection of random forests, which sounds daunting…

My stay in Edinburgh was quite brief as I drove to the Highlands after the seminar, heading to Fort William, Although the weather was rather ghastly, the traffic was fairly light and I managed to get there unscathed, without hitting any of the deer of Rannoch Mor (saw one dead by the side of the road though…) or the snow banks of the narrow roads along Loch Lubnaig. And, as usual, it still was a pleasant feeling to drive through those places associated with climbs and hikes, Crianlarich, Tyndrum, Bridge of Orchy, and Glencoe. And to get in town early enough to enjoy a quick dinner at The Grog & Gruel, reflecting I must have had half a dozen dinners there with friends (or not) over the years. And drinking a great heather ale to them!

the painted man

Another of those fantasy books I bought on the spur of the moment, without prior information, and that I ended liking very much! Indeed, when I was in the UK in April, I bought a few books for my son in Fort William, of all places!, and The painted man by Peter Brett was one of them. My son got very enthusiastic about it and read it within a few days. Then kept asking about the sequel… (Note that the book strangely has an alternative title in the US, The warded man. With the same so-so cover. Why?! Because UK readers could not understand the word warded?! Because US readers would think The painted man was about American indians?!)

So I took the book with me to Guérande to see how good this was and I got hooked as well, finishing it in three days. The sequel, The Desert Spear, is already sitting on top of the to-read-pile! The central theme of the book(s) is a variation on the “fear of the dark” theme, when darkness is full of real dangers, also found in, e.g., Sanderson’s Mistborn series or Terry Goodkind’s (rather annoying) Wizzard’s First Rule or yet Barbara Hambly’s mosy enjoyable (if older) Darwatch trilogy. And, of course, the forerunner H.G. Wells’ Time Machine where morlocks feed upon elois… Not a very promising start, then, especially when the three main characters are three (pre-)teenagers embarking upon their own quest and of course doomed to meet at some point in the story. However, Brett manages to turn this classic in the genre into something different and highly gripping. One of the attractions of the story is that the demons (or corelings) that come out of the ground when the sun sets down are not described into painful details, only their deadly power matters and it seems so overwhelming that the notion of fighting them does not make sense, either to the inhabitants of this universe or to the reader. When one character, Arlen, decides to, nonetheless, it is a major surprise (mild spoiler warning!) that he survives the first night, the first month and then the whole book! Although there are several deus ex machina interventions to make this possible, the story flows rather nicely and Arlen turns into the major character in the book, Further, his growing powers against the night demons come at the price of distancing himself from the other people and (stronger spoiler warning!) getting more similar to the beings he obsessively pursues. Which is why the other characters are destined to meet him. And help him recover his humanity.

Of course, this is not a perfect book. Besides the recurrence of happenstance moments, some characters are too caricatural. For instance, as in several recent fantasy novels (Richard Morgan’s Cold Commands to just pick one!), one desert-related part of the world follows an Islamic-type culture that carries all the clichés about Muslim countries. This makes the book sounds quite ethno-centric, with the bigoted and superstitious but good at heart communities from the North getting the better part over the fanatic, sexists and untrustworthy denizens of the South. I am actually afraid the second volume The Desert Spear will see more of this simplistic opposition as the southern desert tribes start an invasion very much reminding me of the Muslim invasions of the 7th and 8th centuries… The female character, Leesha, is also rather inconsistent, from strong to weak to strong again, a flaw in the story, esp. against the much more coherent Arlen. But, all in all, this remains a terrific first book and many readers seem to have felt the same way from the mostly positive reviews on line. I am eagerly waiting to get my Desert Spear back so that I can read it!

UK trip [#4]

Today I am off (again: 4th time this year) to England, then Scotland, for a week. The first stop is in Bristol, where the Statistics group holds a four day workshop on Confronting Intractability in Statistical Inference, supported by SuSTain. This is a perfect theme for presenting and discussing ABC research, most obviously. I am looking forward to be back in Bristol with friends of many years. From Bristol, I will fly to Glasgow, where I give a seminar at the University of Glasgow on Friday and meet with Mike Titterington. Once more, I am looking forward to be back in Glasgow, having visited the University quite regularly since 1995.

Having a free weekend in Scotland can only imply one thing for me: head for the West and get to the mountains. Depending on the weather and on the (dis)continuation of a dark series with bag transportation, I hope to hike up Bidean nam Bian in Glencoe (as I did in 1995 upon Peter Green’s recommendation!) and climb Ben Nevis by Tower Ridge or even Point Five Gully (the epitome of Scottish Grade V!) if the ice and cold weather hold until Sunday (foolish hope!)…

And from Fort William I will drive to Edinburgh, where Mark Girolami, Antonietta Mira and myself are organizing an ICMS meeting on advances in MCMC on Monday-Wednesday of the following week. There is overlap in themes and participants between this workshop and the one in Bristol, but also a broader spectrum in Edinburgh, incl. a coverage of significant applications. Needless to say, I am also looking forward this meeting, as well as being in Edinburgh and staying in the vicinity of Arthur’s Seat once again!