BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Towards explainable computational biology: small steps and many qu
 estions - Bianca Dumitrascu 
DTSTART:20201026T163000Z
DTEND:20201026T170000Z
UID:TALK153376@talks.cam.ac.uk
CONTACT:Bingqing Cheng
DESCRIPTION:Large\, high capacity\, deep learning algorithms and models\, 
 trained on large amounts of data have shown to achieve impressive performa
 nce and generalize well. As newer\, cheaper types of data querying biologi
 cal systems are currently more available than ever\, the popularity of dee
 p learning in computational biology has increased. However\, with deeper m
 odels comes greater responsibility. Is the complexity of such models warra
 nted and can they bring new insight into scientific decision-making?  In t
 his talk\, I will survey a number of challenges in computational biology\,
  with a  focus on the problem of interpretable feature selection in spatia
 l single cell RNA sequencing data. We will discuss linear approaches to se
 lect relevant features using lasso\, and extensions to the context of deep
  learning using variational autoencoders and the famous gumbel softmax tri
 ck. This is joint work with Nabeel Sarwar and Soledad ViIllar.
LOCATION:virtual ZOOM meeting ID: 263 591 6003\, Passcode: 000042\, https:
 //us02web.zoom.us/j/2635916003?pwd=ZlBEQnRENGwxNmJGMENGMWxjak5nUT09
END:VEVENT
END:VCALENDAR