Archive for University of Oxford

Judith’s colloquium at Warwick

Posted in Statistics with tags , , , , , , , , on February 21, 2020 by xi'an

MCqMC2020 key dates

Posted in pictures, Statistics, Travel, University life with tags , , , , , , , , , on January 23, 2020 by xi'an

A reminder of the key dates for the incoming MCqMC2020 conference this summer in Oxford:

Feb 28, Special sessions/minisymposia submission
Mar 13, Contributed abstracts submission
Mar 27, Acceptance notification
Mar 27, Registration starts
May 8, End of early bird registration
June 12, Speaker registration deadline
Aug 9-14 Conference

and of the list of plenary speakers

Yves Atchadé (Boston University)
Jing Dong (Columbia University)
Pierre L’Ecuyer (Université de Montreal)
Mark Jerrum (Queen Mary University London)
Gerhard Larcher (JKU Linz)
Thomas Muller (NVIDIA)
David Pfau (Google DeepMind)
Claudia Schillings (University of Mannheim)
Mario Ullrich (JKU Linz)

rationality and superstition

Posted in Books with tags , , , , , , , , on December 4, 2019 by xi'an

As I am about to read The Secret Commonwealth, the second volume in his Book of Dust trilogy, I found that Philip Pullman wrote a fairly interesting piece inspired from a visit to an 2018 exhibition at the Ashmolean Museum in Oxford, dedicated to magic and witchcraft. Which I enjoyed reading even though I do not agree with most points. Even though the human tendency to see causes in everything, hidden or even supernatural if need be, explains for superstition and beliefs in magics, the Enlightenment and rise of rationality saw the end of the witch-hunt craze of the 16th and early 17th Centuries (with close to 50,000 executions throughout Europe.

“…rationalism doesn’t make the magical universe go away (…) When it comes to belief in lucky charms, or rings engraved with the names of angels, or talismans with magic squares, it’s impossible to defend it and absurd to attack it on rational grounds because it’s not the kind of material on which reason operates. Reason is the wrong tool. Trying to understand superstition rationally is like trying to pick up something made of wood by using a magnet.”

“Whether witches were “filthy quislings” or harmless village healers, they and those who believed in witchcraft and magic existed in a shared mental framework of hidden influences and meanings, of significances and correspondences, whether angelic, diabolic, or natural (…)  a penumbra of associations, memories, echoes and correspondences that extend far into the unknown. In this way of seeing things, the world is full of tenuous filaments of meaning, and the very worst way of trying to see these shadowy existences is to shine a light on them.”

“I simply can’t agree with (Richard Dawkins’): “We don’t have to invent wildly implausible stories: we have the joy and excitement of real, scientific investigation and discovery to keep our imaginations in line.” (The Magic of Reality, 2011). If we have to keep our imaginations in line, it’s because we don’t trust them not to misbehave. What’s more, only scientific investigation can disclose what’s real. On the contrary, I’d rather say that there are times when we have to keep our reason in line. I daresay that the state of Negative Capability, where imagination rules, is in fact where a good deal of scientific discovery begins. “

Florence Nightingale Bicentennial Fellowship and Tutor in Statistics and Probability in Oxford [call]

Posted in Statistics, Travel, University life with tags , , , , , on July 29, 2019 by xi'an

Reposted: The Department of Statistics is recruiting a Florence Nightingale Bicentennial Fellowship and Tutor in Statistics and Probability with effect from October 2019 or as soon as possible thereafter. The post holder will join the dynamic and collaborative Department of Statistics. The Department carries out world-leading research in applied statistics fields including statistical and population genetics and bioinformatics, as well as core theoretical statistics, computational statistics, machine learning and probability. This is an exciting time for the Department, which relocated to new premises on St Giles’ in the heart of the University of Oxford in 2015. Our newly-renovated building provides state-of-the-art teaching facilities and modern space to facilitate collaboration and integration, creating a highly visible centre for Statistics in Oxford. The successful candidate will hold a doctorate in the field of Statistics, Mathematics or a related subject. They will be an outstanding individual who has the potential to become a leader in their field. The post holder will have the skills and enthusiasm to teach at undergraduate and graduate level, within the Department of Statistics, and to supervise student projects. They will carry out and publish original research within their area of specialisation. We particularly encourage candidates working in areas that link with existing research groups in the department to apply. The deadline for application is September 30, 2019.

If you would like to discuss this post and find out more about joining the academic community in Oxford, please contact Professor Judith Rousseau or Professor Yee Whye Teh. All enquiries will be treated in strict confidence and will not form part of the selection decision.

postdoc position still open

Posted in pictures, Statistics, University life with tags , , , , , , , , , , , , , , on May 30, 2019 by xi'an

The post-doctoral position supported by the ANR funding of our Paris-Saclay-Montpellier research conglomerate on approximate Bayesian inference and computation remains open for the time being. We are more particularly looking for candidates with a strong background in mathematical statistics, esp. Bayesian non-parametrics, towards the analysis of the limiting behaviour of approximate Bayesian inference. Candidates should email me (gmail address: bayesianstatistics) with a detailed vita (CV) and a motivation letter including a research plan. Letters of recommendation may also be emailed to the same address.

selecting summary statistics [a tale of two distances]

Posted in Books, Statistics with tags , , , , , , , , , , , , , , on May 23, 2019 by xi'an

As Jonathan Harrison came to give a seminar in Warwick [which I could not attend], it made me aware of his paper with Ruth Baker on the selection of summaries in ABC. The setting is an ABC-SMC algorithm and it relates with Fearnhead and Prangle (2012), Barnes et al. (2012), our own random forest approach, the neural network version of Papamakarios and Murray (2016), and others. The notion here is to seek the optimal weights of different summary statistics in the tolerance distance, towards a maximization of a distance (Hellinger) between prior and ABC posterior (Wasserstein also comes to mind!). A sort of dual of the least informative prior. Estimated by a k-nearest neighbour version [based on samples from the prior and from the ABC posterior] I had never seen before. I first did not get how this k-nearest neighbour distance could be optimised in the weights since the posterior sample was already generated and (SMC) weighted, but the ABC sample can be modified by changing the [tolerance] distance weights and the resulting Hellinger distance optimised this way. (There are two distances involved, in case the above description is too murky!)

“We successfully obtain an informative unbiased posterior.”

The paper spends a significant while in demonstrating that the k-nearest neighbour estimator converges and much less on the optimisation procedure itself, which seems like a real challenge to me when facing a large number of particles and a high enough dimension (in the number of statistics). (In the examples, the size of the summary is 1 (where does the weight matter?), 32, 96, 64, with 5 10⁴, 5 10⁴, 5 10³ and…10 particles, respectively.) The authors address the issue, though, albeit briefly, by mentioning that, for the same overall computation time, the adaptive weight ABC is indeed further from the prior than a regular ABC with uniform weights [rather than weighted by the precisions]. They also argue that down-weighting some components is akin to selecting a subset of summaries, but I beg to disagree with this statement as the weights are never exactly zero, as far as I can see, hence failing to fight the curse of dimensionality. Some LASSO version could implement this feature.

congrats, Prof Rousseau!

Posted in Statistics with tags , , , , , , , , on April 4, 2019 by xi'an