BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Neural Networks with Euclidean Symmetry for Physical Sciences - Te
 ss E. Smidt\, Lawrence Berkeley National Laboratory
DTSTART:20201109T163000Z
DTEND:20201109T170000Z
UID:TALK153739@talks.cam.ac.uk
CONTACT:Bingqing Cheng
DESCRIPTION:Atomic systems (molecules\, crystals\, proteins\, nanoclusters
 \, etc.) are naturally represented by a set of coordinates in 3D space lab
 eled by atom type. This is a challenging representation to use for neural 
 networks because the coordinates are sensitive to 3D rotations and transla
 tions and there is no canonical orientation or position for these systems.
  One of the motivations for incorporating symmetry into machine learning m
 odels on 3D data is to eliminate the need for data augmentation -- the 500
 -fold increase in brute-force training necessary for a model to learn 3D p
 atterns in arbitrary orientations. \n\nMost symmetry-aware machine learni
 ng models in the physical sciences avoid augmentation through invariance\,
  throwing away coordinate systems altogether. But this comes at a price\; 
 many of the rich consequences of Euclidean symmetry are lost: geometric te
 nsors\, point groups\, space groups\, degeneracy\, atomic orbitals\, etc.\
 n\nWe present a general neural network architecture that faithfully treats
  the equivariance of physical systems and naturally handles 3D geometry an
 d operates on the scalar\, vector\, and tensor fields that characterize th
 em. Our networks are locally equivariant to 3D rotations and translations 
 at every layer. In this talk\, we describe how the networks achieves these
  equivariances and demonstrate the capabilities of our network using simpl
 e tasks.  We provide concrete coding examples for how to build these model
 s using e3nn: a modular open-source PyTorch framework for Euclidean Neural
  Networks (https://e3nn.org).
LOCATION:virtual ZOOM meeting ID: 263 591 6003\, Passcode: 000042\, https:
 //us02web.zoom.us/j/2635916003?pwd=ZlBEQnRENGwxNmJGMENGMWxjak5nUT09
END:VEVENT
END:VCALENDAR