BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Deep learning for cosmology: parameter measurement and generation 
 of simulations
DTSTART:20190417T110000
DTEND:20190417T120000
DTSTAMP:20260407T045455Z
UID:7958c26948a44f9713229845ec7307de57e8e0545e52c0464c27f5a8
CATEGORIES:Public Science Events
DESCRIPTION:Dr Tomasz Kacprzak\nDeep learning-based analysis methods are g
 aining interest in cosmology due to their unique ability to create very ri
 ch and complex models.\nThese models are particularly well suited for anal
 ysis of large scale structure data\, as the matter density fields are comp
 rised of highly nonlinear\, complicated features\, such as halos\, filamen
 ts\, sheets and voids.\nBut can this information be utilised by the deep l
 earning algorithm to gain a better understanding of the cosmological model
 ?\nIn this talk I will present the application of Convolutional Neural Net
 works (CNNs) for constraining cosmological parameters.\nI will compare the
  constraining power against the commonly used statistic\, the power spectr
 um\, and explore different regimes in quality of data and simulations.\nFi
 nally\, I will introduce the Generative Adversarial Networks (GANs): a CNN
 -based technique\, which can learn from a training set and then generate n
 ew\, statistically similar data.\nI will present a study of applying GANs 
 to generating samples of the cosmic web and discuss the prospects of apply
 ing them to render 2D and 3D N-body simulation - like data.\n 
LOCATION:Martigny\, room 106 https://www.google.com/maps/d/viewer?mid=1Wi7
 CoK1Apqmnxt-dLMI8FpRgE0s&msa=0&ll=46.108789091068445%2C7.082373000000075&s
 pn=0.006337%2C0.018915&z=16
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR