BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Infusing Structure and Knowledge into Biomedical AI Algorithms - M
 arinka Zitnik (Harvard University)
DTSTART:20211123T140000Z
DTEND:20211123T150000Z
UID:TALK164860@talks.cam.ac.uk
DESCRIPTION:Grand challenges in biology and medicine often lack annotated 
 examples and require generalization to entirely new scenarios not seen dur
 ing training. However\, standard supervised learning is incredibly limited
  in scenarios\, such as designing novel medicines\, modeling emerging path
 ogens\, and treating rare diseases. In this talk\, I present our efforts t
 o overcome these obstacles by infusing structure and knowledge into learni
 ng algorithms. First\, I outline our subgraph neural networks that can dis
 entangle distinct aspects of subgraph topology. I then present a general-p
 urpose approach for few-shot learning on graphs. At the core is the notion
  of local subgraphs that transfer knowledge from one task to another\, eve
 n when only a handful of labeled examples are available. This principle is
  theoretically justified as we show the evidence for predictions can be fo
 und in subgraphs surrounding the targets. I conclude with applications in 
 drug development and precision medicine where the algorithmic predictions 
 were validated in human cells and led to the discovery of a new class of d
 rugs.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR