BEGIN:VCALENDAR
PRODID:-//Columba Systems Ltd//NONSGML CPNG/SpringViewer/ICal Output/3.3-
 M3//EN
VERSION:2.0
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
DTSTAMP:20220323T144649Z
DTSTART:20220329T130000Z
DTEND:20220329T140000Z
SUMMARY:Turing Fellow 'Spotlight':29 March (rescheduled)
UID:{http://www.columbasystems.com/customers/uom/gpp/eventid/}n1fx-l0v59p
 5p-gf53x1
DESCRIPTION:*Please note: this event was originally scheduled to take pla
 ce on 1 March but has been rescheduled due to the UCU strike action.*\n\
 nThe Digital Futures team are hosting a series of online talks by The Un
 iversity of Manchester's new Turing Fellows for 2021-22.\n\nFor the fina
 l event in this series\, new Turing Fellow Christopher Conselice is join
 ed by Turing AI Fellow Anna Scaife.\n\nChristopher Conselice\, Professor
  of Extragalactic Astronomy.\n\nChristopher's research is focused on gal
 axy formation and evolution through cosmic time. He works with space and
  ground-based data and computer simulations to determine how galaxies ha
 ve formed from the earliest epochs in the universe until today. He has l
 ed several large HST surveys including the GOODS NICMOS Survey (GNS) and
  is a founding member of other large surveys including CANDELS and GOODS
 .\n\nChristopher's talk is titled: 'Machine learning applications in ast
 rophysics: Large surveys and galaxy properties/evolution'\n\nRecently th
 e application of machine learning\, especially deep learning\, to astron
 omy and astrophysics has become very popular. Open use of large data set
 s in astronomy makes it ideal for applications of machine learning\, yet
  many of the tools\, procedures\, and adaptability are still in their ea
 rly stages. I will present some of our work on these topics including us
 ing supervised\, unsupervised\, and regression analyses for determining 
 galaxy properties and evolution. I will also discuss upcoming astronomy 
 missions which will contain hundreds of millions of galaxies\, stars\, q
 uasars\, defects and other detected objects that can only be classified 
 and studied using machine learning techniques that are in the process of
  being developed.\n\nAnna Scaife\, Professor of Radio Astronomy at Jodre
 ll Bank Centre for Astrophysics.\n\nAnna’s Turing AI Fellowship focuses 
 on AI for discovery in data intensive astrophysics. In this era of big d
 ata astrophysics\, the use of machine learning to extract scientific inf
 ormation is essential to successfully utilise facilities such as the Squ
 are Kilometre Array (SKA) telescopes. These telescopes have data rates s
 o large that the raw data cannot be stored and even using the compressed
  data products will require a super-computer.\n\nAnna's talk is titled '
 AI in the SKA Era: Challenges for Bayesian Neural Networks in Radio Gala
 xy Classification'.\n\nThe expected volume of data from the new generati
 on of scientific facilities such as the Square Kilometre Array (SKA) rad
 io telescope has motivated the expanded use of semi-automatic and automa
 tic machine learning algorithms for scientific discovery in astronomy. I
 n this field\, the robust and systematic use of machine learning faces a
  number of specific challenges including a paucity of labelled data for 
 training (paradoxically\, although we have too much data\, we don't have
  enough)\, a clear understanding of the effect of biases introduced due 
 to observational and intrinsic astrophysical selection effects in the tr
 aining data\, and motivating a quantitative statistical representation o
 f outcomes from decisive AI applications. In this seminar I will talk sp
 ecifically about the challenge of recovering well-calibrated uncertainti
 es from Bayesian neural networks when classifying radio galaxies\, a can
 onical example of a radio astronomy AI application. I will discuss how b
 oth model and likelihood misspecification can affect this calibration\, 
 how these effects potentially contribute to the cold posterior effect se
 en when building models using real astronomical data and what steps we c
 an take to address these problems.
STATUS:TENTATIVE
TRANSP:TRANSPARENT
CLASS:PUBLIC
END:VEVENT
END:VCALENDAR