BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Neural Networks for High-Dimensional Tabular Biomedical Datasets -
  Andrei Margeloiu (University of Cambridge)
DTSTART:20230221T130000Z
DTEND:20230221T140000Z
UID:TALK195256@talks.cam.ac.uk
CONTACT:Mateja Jamnik
DESCRIPTION:Modern machine learning algorithms frequently overfit on small
 -sample size and high-dimensional tabular datasets\, which are common in m
 edicine\, bioinformatics and drug discovery. How can we reduce the overfit
 ting on tabular datasets with D>>N?\n\nThis talk presents two neural metho
 ds for learning from small-sample size and high-dimensional tabular datase
 ts. First\, we present WPFS\, a parameter-efficient neural architecture th
 at performs global feature selection during training. Second\, we present 
 GCondNet\, a general approach which combines Graph Neural Networks (GNNs) 
 for incorporating the implicit relationships between samples when training
  standard neural networks. GCondNet exploits the high-dimensionality of th
 e data by creating many small graphs to capture the structure between samp
 les within a feature. We show that WPFS and GCondNet outperform both stand
 ard and more recent methods on real-world biomedical datasets.\n\n"You can
  also join us on Zoom":https://zoom.us/j/99166955895?pwd=SzI0M3pMVEkvNmw3Q
 0dqNDVRalZvdz09\n
LOCATION:Lecture Theatre 2
END:VEVENT
END:VCALENDAR