BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Ranking the information content of distance measures through the i
 nformation imbalance - Aldo Glielmo ( International School for Advanced St
 udies (SISSA)
DTSTART:20210628T153000Z
DTEND:20210628T163000Z
UID:TALK161194@talks.cam.ac.uk
CONTACT:Bingqing Cheng
DESCRIPTION:Real-world data typically contain a large number of features t
 hat are often heterogeneous in nature\, relevance\, and also units of meas
 ure.  When assessing the similarity between data points\, one can build va
 rious distance measures using subsets of these features. Using the fewest 
 features but still retaining sufficient information about the system is cr
 ucial in many statistical learning schemes\, particularly when data are sp
 arse. \nIn my talk I will describe the “information imbalance”: a nove
 l statistical concept that quantifies the relative information retained wh
 en using two different distance measures\, and determine if they are equiv
 alent\, independent\, or if one is more informative than the other. \nI wi
 ll then show how the information imbalance can be used to find the most in
 formative distance measure out of a pool of candidates\, and present appli
 cations of this idea for the analysis of the Covid-19 epidemic spreading a
 s well as for the construction of optimally informative descriptors of phy
 sical systems.
LOCATION:virtual ZOOM meeting ID: 263 591 6003\, Passcode: 000042\, https:
 //us02web.zoom.us/j/2635916003?pwd=ZlBEQnRENGwxNmJGMENGMWxjak5nUT09
END:VEVENT
END:VCALENDAR