BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:A Bayesian method to localize lost gamma sources
DTSTART:20190507T092000Z
DTEND:20190507T094000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9249@indico.ess.eu
DESCRIPTION:Speakers: Antanas Bukartas (Lund University)\, Robert Finck (L
und University)\, Jonas Wallin (Lund University)\, Christopher Rääf (Lun
d University)\n\nRadioactive sources can sometimes be lost or misplaced de
spite the existing rigorous safety rules. Lost sources must be found as so
on as possible to avoid inflicting harm to the public. Regardless of the t
ype of equipment used it is desirable to use as much information as possib
le from the measurements to draw conclusions about the activity and locati
on of a detected source. Using Bayesian inference it is possible to obtain
a probability distribution for the position and activity of an unshielded
gamma source in one pass with a mobile gamma spectrometry vehicle.\nThe a
im of this research was to investigate the feasibility of a Bayesian algor
ithm for mobile gamma spectrometry\,test its accuracy in determining the l
ocation and activity.\n\nhttps://indico.ess.eu/event/1191/contributions/92
49/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9249/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Spatio-Temporal Reconstructions of Global CO2-Fluxes using Gaussia
n Markov Random Fields
DTSTART:20190507T094000Z
DTEND:20190507T100000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9250@indico.ess.eu
DESCRIPTION:Speakers: Unn Dahlén\, Marko Scholze (Lunds Universitet)\, Jo
han Lindström (Lunds Universitet)\n\nAtmosperic invere modelling is a met
hod for constraining Earth surface fluxes (sinks and sources) of green hou
se gases using measurements of athmosperic concentrations. The (linear) li
nk between atmospheric concentration and fluxes are provided by an atmosph
eric transport model. Since the number of unknown surfaces fluxes is much
larger than the number of observed atmospheric concentrations\, the invers
e problem is ill-conditioned. Requiring further assumption on the fluxes\,
leading to a Bayesian model. Hitorically\, fluxes are discretized to a gr
id and modelled by a multivariate Gaussian distribution. Instead\, we defi
ne the flux on a continuous spatial domain\, with fluxes modelled as Gauss
ian Markov Random Fields\, including both spatial and temporal dependence.
\n\nhttps://indico.ess.eu/event/1191/contributions/9250/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9250/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Extending Bayes to Make Optimal Decisions
DTSTART:20190507T120000Z
DTEND:20190507T122000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9248@indico.ess.eu
DESCRIPTION:Speakers: Jonas Kristoffer Lindeløv (Department of Communicat
ion and Psychology\, Aalborg University\, Denmark)\n\nUtility Theory allow
you to make optimal decisions in the face of uncertainty. For example\, w
hat bidding price would maximize your earnings\, taking the chance of fail
ure into account? Utility Theory latches nicely onto Bayesian Inference. O
nce you have a posterior distribution\, you need only a few more lines of
code to apply a utility function (aka loss function) and identify the deci
sion that optimizes said utility. This approach scales well to more comple
x models and decisions. We will use R and rstanarm/brms for Bayesian infer
ence and hand-code the utility. An R notebook with worked examples will ac
company the tutorial.\n\nhttps://indico.ess.eu/event/1191/contributions/92
48/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9248/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Visualisation for refining and communicating Bayesian analyses
DTSTART:20190507T104500Z
DTEND:20190507T113000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9321@indico.ess.eu
DESCRIPTION:Speakers: Robert Grant (BayesCamp)\n\nIntroductory statistics
classes teach the value of basic plots to diagnose problems with regressio
n models. Like most of us\, I found these very boring and spent a long tim
e avoiding them. Now that I focus on complex Bayesian models\, I find visu
alisation of variables\, predictions and residuals more useful than ever.
Analysts need to track down problems where the model is not fitting the da
ta well\, and also where the sampling algorithm is performing poorly\, whe
re the priors may be a bad choice\, and where the likelihood may be a bad
choice. In addition to this\, Bayesian models fitted with simulation are f
ertile ground for visual communication of findings. In this talk\, I will
review relevant general principles of effective visualisation\, recent wor
k on Bayesian workflow\, and the role that interactive graphics have to pl
ay.\n\nhttps://indico.ess.eu/event/1191/contributions/9321/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9321/
END:VEVENT
BEGIN:VEVENT
SUMMARY:How to deal with a noisy zero – a simple Bayesian treatment for
small angle neutron scattering
DTSTART:20190507T133000Z
DTEND:20190507T135000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9255@indico.ess.eu
DESCRIPTION:Speakers: Alexander Holmes (European Spallation Source ERIC)\n
\nNeutron scattering measurements are an ideal case for Bayesian analysis
– statistics are limited\, measurement time is expensive and there is of
ten relevant background information.\n\nI will present an example of small
angle neutron scattering from superconducting vortex lattices. Most of th
e signal detected is irrelevant\, and contributes nothing but noise to the
final result. A Bayesian treatment of the results\, allows a huge enhanc
ement of the signal to noise ratio\, and treats missing data sensibly.\n\n
I will further argue that Bayesian analysis offers a huge amount to the sc
attering community and would benefit significantly from more systematic su
pport from institutions such as the ESS. \n\n[A. T. Holmes\, Phys. Rev. B
90\, 024514][1]\n\n[1]: https://doi.org/10.1103/PhysRevB.90.024514\n\nhttp
s://indico.ess.eu/event/1191/contributions/9255/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9255/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Automatic Learning of Summary Statistics for Approximate Bayesian
Computation Using Deep Learning
DTSTART:20190507T143000Z
DTEND:20190507T145000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9254@indico.ess.eu
DESCRIPTION:Speakers: Samuel Wiqvist (Lund University)\, Umberto Picchini
(Departmentof Mathematical Sciences\, Chalmers University of Technologyan
d the University of Gothenburg)\, Jes Frellsen (IT University of Copenhag
en)\, Pierre-Alexandre Mattei (IT University of Copenhagen)\n\nLearning s
ummary statistics is a fundamental problem in Approximate Bayesian Computa
tion (ABC). The problem of learning summary statistics is in fact the main
challenge when applying ABC in practice\, and affects the resulting infer
ence considerably. Deep learning methods have previously been used to lear
n summary statistics for ABC. Here we introduce a novel deep learning arch
itecture (Partially Exchangeable Networks\, PENs)\, with the purpose of *a
utomatically* learning summary statistics for ABC. Our case studies show t
hat our methodology outperforms other popular methods\, resulting in more
accurate ABC inference for Markovian data.\n\nJoint work with Pierre-Alexa
ndre Mattei\, Umberto Picchini and Jes Frellsen.\n\nhttps://indico.ess.eu/
event/1191/contributions/9254/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9254/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Rich-man's Monte Carlo: Uncertainty Analysis in Excel
DTSTART:20190507T122000Z
DTEND:20190507T124000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9251@indico.ess.eu
DESCRIPTION:Speakers: Dmytro Perepolkin (Lund University)\n\nAdoption of U
ncertainty Analysis in modern business environment is often challenging du
e to gaps in relevant skills and tooling (especially among decision makers
). At the same time Excel is ubiquitous and can be used to build decision
maker's intuition about uncertainties. This talk will introduce typical bu
siness problem faced by businesses and organizations on daily basis and sh
owcase a fully transparent formula-only Excel model with no dependency on
external libraries or macros\, that can enlighten and inform about effect
of uncertainties on business outcome(s).\n\nIn this talk we will discuss f
raming\, expert knowledge elicitation\, modeling\, visualization and commu
nication of results to facilitate discussion about uncertainties and ultim
ately aid the decision making.\n\nhttps://indico.ess.eu/event/1191/contrib
utions/9251/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9251/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Introducing Bayesian Stats through Signal Detection Theory
DTSTART:20190507T082000Z
DTEND:20190507T084000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9309@indico.ess.eu
DESCRIPTION:Speakers: Gerit Pfuhl (UiT)\n\nTo avoid starting with a formul
a a detour via utilized Signal detection theory (uSDT) familiarizes psycho
logy undergraduates with some of the basic concepts in Bayesian statistics
. uSDT includes payoffs (utility functions)\, base rates\, and varies simi
larities\, illustrated on perceptual decision processes. Payoffs and base
rates influences bias\, assisting students in understanding models and pri
ors in Bayesian statistics.\n\nhttps://indico.ess.eu/event/1191/contributi
ons/9309/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9309/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Prior thoughts on mixed-membership models in linguistics
DTSTART:20190507T090000Z
DTEND:20190507T092000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9311@indico.ess.eu
DESCRIPTION:Speakers: Chundra Cathcart\, Gerd Carling\n\nBayesian mixed-me
mbership models are popular in linguistics\, as they explicitly model cont
act between languages (Reesink et al 2009\, Syrjänen et al 2016). Most li
nguistic applications use the biological Structure program (Pritchard et a
l 2000) with default presets\, fixing the concentration parameter of the p
opulation-level Dirichlet prior over allele frequency (treated as an analo
g for the language-level prior over features) at 1. We show\, using a cros
slinguistic typological database\, that there are linguistically meaningfu
l consequences for the choice of this hyperparameter (either fixed at diff
erent values\, or inferred from the data) using a series of posterior pred
ictive checks designed for mixed-membership models (Mimno et al 2015).\n\n
https://indico.ess.eu/event/1191/contributions/9311/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9311/
END:VEVENT
BEGIN:VEVENT
SUMMARY:What cause successful learning in Bayesian methods?
DTSTART:20190507T080000Z
DTEND:20190507T082000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9252@indico.ess.eu
DESCRIPTION:Speakers: George Moroz (National Research University Higher Sc
hool of Economics)\n\nI would like to share the comparison of three groups
of students that I taught Bayesian methods this year:\n\n - mixed group (
psychology\, biology...) with good background in R and frequentist statist
ics\;\n - linguistic group (3d year) with medium background in R and frequ
entist statistics\;\n - further education group in Computer Linguistic wit
h beginner R and no statistical background.\n\nI expected first and second
groups to be more successful than the third one\, but the shocking result
was that the third and the first groups were more successful than the sec
ond one. I will try to explain the obtained result.\n\nhttps://indico.ess.
eu/event/1191/contributions/9252/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9252/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayesian vs. Frequentism for experimentalists
DTSTART:20190507T113000Z
DTEND:20190507T115000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9253@indico.ess.eu
DESCRIPTION:Speakers: Jakob Lavröd (IYPT Sweden)\n\nA common rebuttal to
Bayesian methods is that they are appropriate for large and complex proble
ms (containing prior information and hidden variables)\, and most undergra
duate teaching is often based upon frequentist methods. Starting from the
context of the practical experimentalist\, we explore the difference betwe
en the Bayesian and Frequentist methodology and highlights the advantages
of teaching the Bayesian perspective already from the start. Many of the e
xamples and the material come from the introductory course given for the s
tudents enrolled in the high school giftedness program “International Yo
ung Physicists’ Tournament”\, an attempt at introducing Bayesian think
ing into experimental practice.\n\nhttps://indico.ess.eu/event/1191/contri
butions/9253/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9253/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayesian inference of conformational ensembles from small-angle sc
attering data
DTSTART:20190507T135000Z
DTEND:20190507T141000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9308@indico.ess.eu
DESCRIPTION:Speakers: Wojciech Potrzebowski (European Spallation Source\,
ERIC)\, Ingemar Andre (Biochemistry and Structural Biology\, LU)\n\nSmall-
angle scattering (SAS) uses x-ray or neutron scattering at small angles to
investigate the structure of materials at the scale about 1-100nm. SAS is
uniquely suited to study the conformational ensembles adopted by multidom
ain proteins. However\, analysis is complicated by the limited information
content in SAS data and care must be taken to avoid constructing overly c
omplex ensemble models and fitting to noise in the experimental data. To a
ddress these challenges\, we developed a method based on Bayesian statisti
cs that infers conformational ensembles from a structural library generate
d by all-atom Monte Carlo simulations. The method involves a fast model se
lection based on variational Bayesian inference that maximizes the model e
vidence\, followed by a complete Bayesian inference of population weights.
\n\nhttps://indico.ess.eu/event/1191/contributions/9308/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9308/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayesian determination of the effect of a deep eutectic solvent on
the structure of lipid monolayers
DTSTART:20190507T141000Z
DTEND:20190507T143000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9310@indico.ess.eu
DESCRIPTION:Speakers: Andrew McCluskey (Diamond Light Source & University
of Bath)\, Andrew Jackson (European Spallation Source)\, Richard Campbell
(University of Manchester)\, Karen Edler (University of Bath)\, Tom Arnold
(European Spallation Source)\, Adrian Sanchez-Fernandez (Lund University)
\, Stephen Parker (University of Bath)\n\nWe present a unique insight from
Bayesian-driven modelling for a series of lipid monolayers at the air-dee
p eutectic solvent (DES) interface using reflectometry measurements.\nA ch
emically-consistent modelling approach shows that the lipid monolayers at
the air-DES interface are similar to those on water\, while removing the n
eed for water-specific constraints.\nFurthermore\, the use of Markov-chain
Monte Carlo sampling enables the quantification of inverse uncertainties
and parameter correlations in the modelling approach.\nFinally\, we discus
s limitations present in the use of Bayesian methods for reflectometry ana
lysis\, and outline future work that will be conducted to overcome these.\
n\nhttps://indico.ess.eu/event/1191/contributions/9310/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9310/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Hierarchical models and their applications in astronomy
DTSTART:20190507T070500Z
DTEND:20190507T075500Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9319@indico.ess.eu
DESCRIPTION:Speakers: Maggie Lieu (ESA)\n\nMaggie is an astrophysics resea
rch fellow working at the European Space Agency in Madrid. Her main resear
ch involves modelling the mass distribution of clusters of galaxies to und
erstand the nature of dark matter and dark energy in our Universe. Maggie
will be talking about how hierarchical models can be a powerful tool for i
nference.\n\nhttps://indico.ess.eu/event/1191/contributions/9319/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9319/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayes@Lund 2019 registration
DTSTART:20190507T063000Z
DTEND:20190507T070000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9323@indico.ess.eu
DESCRIPTION:https://indico.ess.eu/event/1191/contributions/9323/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9323/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Welcome
DTSTART:20190507T070000Z
DTEND:20190507T070500Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9324@indico.ess.eu
DESCRIPTION:Speakers: Alexander Holmes (European Spallation Source ERIC)\n
\nhttps://indico.ess.eu/event/1191/contributions/9324/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9324/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Closing remarks
DTSTART:20190507T145000Z
DTEND:20190507T150000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9322@indico.ess.eu
DESCRIPTION:Speakers: Ullrika Sahlin (Lund University Centre of Environmen
tal and Climate Research)\, Rasmus Bååth (Lund University\; King Enterta
inment)\n\nhttps://indico.ess.eu/event/1191/contributions/9322/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9322/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayesian Deep Learning Applications in Biomedicine
DTSTART:20190507T124000Z
DTEND:20190507T130000Z
DTSTAMP:20240622T194100Z
UID:indico-contribution-9442@indico.ess.eu
DESCRIPTION:Speakers: Nikolay Oskolkov (Lund University\, Department of Bi
ology)\n\nNext Generation Sequencing technologies gave rise to manifolds o
f Biomedical Big Data which is particularly manifested in the area of sing
le cell transcriptomics where millions of cells are sequenced. Deep Learni
ng (DL) is an ideal framework for analyzing large amounts of data and buil
ding predictive models for Clinical Diagnostics within the concept of Prec
ision Medicine. Bayesian DL adds an important level of patient safety prov
iding uncertainties to the biomedical predictions. Here\, using single-cel
l transcriptomics data I demonstrate how Bayesian DL improves the accuracy
of discovering novel cell sub-populations and dramatically outperforms cl
assical methods when handling unknown cell sub-types.\n\nhttps://indico.es
s.eu/event/1191/contributions/9442/
LOCATION:Palaestra\, Lund University\, Sweden
URL:https://indico.ess.eu/event/1191/contributions/9442/
END:VEVENT
END:VCALENDAR