BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Learning on Graphs with Missing Node Features - Emanuele Rossi\, T
 witter & Imperial College
DTSTART:20211123T131500Z
DTEND:20211123T141500Z
UID:TALK165859@talks.cam.ac.uk
CONTACT:Mateja Jamnik
DESCRIPTION:"Join us on Zoom":https://zoom.us/j/99166955895?pwd=SzI0M3pMVE
 kvNmw3Q0dqNDVRalZvdz09\n\nWhile Graph Neural Networks (GNNs) have recently
  become the de facto standard for modeling relational data\, they impose a
  strong assumption on the availability of the node or edge features of the
  graph. In many real-world applications\, however\, features are only part
 ially available\; for example\, in social networks\, age and gender are av
 ailable only for a small subset of users. We present a general approach fo
 r handling missing features in graph machine learning applications that is
  based on minimization of the Dirichlet energy and leads to a diffusion-ty
 pe differential equation on the graph. The discretization of this equation
  produces a simple\, fast and scalable algorithm which we call Feature Pro
 pagation. We experimentally show that the proposed approach outperforms pr
 evious methods on six common node-classification benchmarks and can withst
 and surprisingly high rates of missing features: on average we observe onl
 y around 4% relative accuracy drop when 99% of the features are missing. M
 oreover\, it takes only 10 seconds to run on a graph with ∼2.5M nodes an
 d ∼123M edges on a single GPU.\n\nBIO: Emanuele is a Machine Learning Re
 searcher at Twitter and a Ph.D. student at Imperial College London\, worki
 ng on Graph Neural Networks and supervised by Prof. Michael Bronstein. His
  research interests span a wide array of topics around graph neural networ
 ks\, including scalability\, dynamic graphs\, and learning with missing no
 de features. Before his current position\, Emanuele was working at Fabula 
 AI\, which was then acquired by Twitter in June 2019. Previously\, he comp
 leted an MPhil at the University of Cambridge and a BEng at Imperial Colle
 ge London\, both in Computer Science.
LOCATION:Zoom
END:VEVENT
END:VCALENDAR