BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Extreme Statistics: Modern Solutions and Open Challenges in the Mo
 deling and Inference of Complex High-Impact Events
DTSTART:20230210T133000
DTEND:20230210T143000
DTSTAMP:20260414T071536Z
UID:c9b7b3a2fdbfe9e4098a66ab2ad8f0c0eac0555416a955d2256baa13
CATEGORIES:Conferences - Seminars
DESCRIPTION:Raphaël HUSER\, KAUST\nSeminar in Mathematics\nAbstract: Rare
 \, low-probability events often lead to the biggest impacts. Therefore\, t
 he development of cutting-edge statistical machine learning approaches for
  modeling\, predicting and quantifying risks is of utmost importance in a 
 variety of fields of applications. Climate scientists\, insurers\, enginee
 rs\, and financial analysts have indeed realized that the greatest risks a
 re often caused by changes in the intensity\, frequency\, extent\, and per
 sistence of extreme events\, rather than changes in their average behavior
 . However\, while datasets are often massive in modern-day applications\, 
 extreme events are always scarce by nature. This makes it doubly challengi
 ng: models needed to realistically describe the dependence structure of hi
 gh-dimensional random vectors are becoming increasingly complex\, but the 
 number of independent replicates used to fit these models remains relative
 ly small. It is thus very important to design specialized models with stro
 ng theoretical support for reliable extrapolation to yet-unseen risk level
 s\, as well as efficient inference methods to fit them to big datasets. In
  this colloquium\, I will first provide a quick overview of classical mode
 ls for spatial extremes and their severe methodological and computational 
 limitations. I will then describe recent progress we have made to develop 
 novel sub-asymptotic spatial models with an improved flexibility in their 
 tail structure. To overcome computational challenges when performing infer
 ence\, I will then discuss neural Bayes estimators\, which are general\, a
 mortized\, likelihood-free estimators constructed from permutation-invaria
 nt neural networks\, that we recently developed for near-optimal and fast 
 inference in complex models. Post-training\, these estimators allow estima
 tion of model parameters and uncertainty quantification in just a fraction
  of second. I will highlight their interesting decision-theoretic connecti
 ons to conventional estimators\, and briefly discuss our on-going work in 
 that area. I will then conclude the talk with some remarks on future resea
 rch directions that have the potential to make an impact across statistics
  as a whole.\n 
LOCATION:MA A1 10 https://plan.epfl.ch/?room==MA%20A1%2010 https://epfl.zo
 om.us/j/66042794625
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR