BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:Accelerating Bayesian Inference and Data Acquisiti
 on via Amortization - Daolang Huang (Aalto Univers
 ity)
DTSTART;TZID=Europe/London:20250624T111500
DTEND;TZID=Europe/London:20250624T121500
UID:TALK232210AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/232210
DESCRIPTION:Many critical applications demand systems that can
  both strategically acquire the most informative d
 ata and instantaneously perform inference based up
 on it. Bayesian inference and Bayesian experimenta
 l design offer principled mathematical means for r
 easoning under uncertainty and for strategically g
 athering data\, respectively. While foundational\,
  both methods introduce notorious computational ch
 allenges. In recent years\, amortized solutions ha
 ve been proposed to address these issues by pre-tr
 aining neural networks\, significantly reducing co
 mputational costs at deployment.\nIn this talk\, I
  will first introduce the Amortized Conditioning E
 ngine (ACE)\, a flexible amortized inference frame
 work that affords conditioning on both observed da
 ta and interpretable latent variables\, the inclus
 ion of priors at runtime\, and outputs predictive 
 distributions for both discrete and continuous dat
 a and latents. I will then share our latest work\,
  the Amortized Active Learning and Inference Engin
 e (ALINE). ALINE combines the advantages of amorti
 zed inference and experimental design into a singl
 e\, unified framework. It is capable of rapidly pr
 oposing valuable data points while simultaneously 
 performing fast\, flexible inference based on the 
 collected data\, thus seamlessly closing the loop 
 between active data acquisition and real-time reas
 oning.
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:
END:VEVENT
END:VCALENDAR