BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:motifNet: Deep learning for system identification of regulatory ne
 tworks - Vincent Fortuin (ETH)
DTSTART:20170614T103000Z
DTEND:20170614T111500Z
UID:TALK73064@talks.cam.ac.uk
CONTACT:Daniel McNamee
DESCRIPTION:System identification of gene regulatory networks based on pro
 teomics or transcriptomics time course data remains a hard problem\, where
  even state-of-the-art algorithms perform relatively poorly in the regime 
 of large system sizes. This work investigates whether machine learning met
 hods\, particularly deep learning\, can exploit features of biological net
 works beyond first principles of chemical kinetics in order to improve per
 formance on this task. We devised a deep neural network architecture calle
 d motifNet\, combining convolutional and recurrent neural network elements
 \, to tackle this problem. Our framework performs better than the state-of
 -the-art comparison methods in terms of area under the ROC and PR curve on
  unseen time course data simulated from the S. cerevisiae and E. coli gene
  regulatory network.
LOCATION:Cambridge University Engineering Department\, CBL\, BE-438 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR