TY - GEN
T1 - Graph Neural Networks for IceCube Signal Classification
AU - Choma, Nicholas
AU - Monti, Federico
AU - Gerhardt, Lisa
AU - Palczewski, Tomasz
AU - Ronaghi, Zahra
AU - Prabhat, Prabhat
AU - Bhimji, Wahid
AU - Bronstein, Michael
AU - Klein, Spencer
AU - Bruna, Joan
N1 - Funding Information:
This work was supported in part by the National Science Foundation under grant number PHY-1307472
Funding Information:
We thank the IceCube Collaboration for their support on this project. This work was supported in part by the National Science Foundation under grant number PHY-1307472 and the U.S. Department of Energy under contract number No. DE-AC02-05CH11231.This research used computational and storage resources of the National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science User Facility. FM and MB are supported in part by ERC Consolidator Grant No. 724228 (LEMAN), Google Faculty Research Awards, an Amazon AWS Machine Learning Research grant, an Nvidia equipment grant, a Radcliffe Fellowship at the Institute for AdvancedStudy, Harvard University, and a Rudolf Diesel Industrial Fellowship at IAS TU Munich. JB and NC are partially supported by the Intel IPCC Program and by the Alfred P. Sloan Foundation.
Publisher Copyright:
© 2018 IEEE.
PY - 2018/7/2
Y1 - 2018/7/2
N2 - Tasks involving the analysis of geometric (graph-and manifold-structured) data have recently gained prominence in the machine learning community, giving birth to a rapidly developing field of geometric deep learning. In this work, we leverage graph neural networks to improve signal detection in the IceCube neutrino observatory. The IceCube detector array is modeled as a graph, where vertices are sensors and edges are a learned function of the sensors spatial coordinates. As only a subset of IceCubes sensors is active during a given observation, we note the adaptive nature of our GNN, wherein computation is restricted to the input signal support. We demonstrate the effectiveness of our GNN architecture on a task classifying IceCube events, where it outperforms both a traditional physics-based method as well as classical 3D convolution neural networks.
AB - Tasks involving the analysis of geometric (graph-and manifold-structured) data have recently gained prominence in the machine learning community, giving birth to a rapidly developing field of geometric deep learning. In this work, we leverage graph neural networks to improve signal detection in the IceCube neutrino observatory. The IceCube detector array is modeled as a graph, where vertices are sensors and edges are a learned function of the sensors spatial coordinates. As only a subset of IceCubes sensors is active during a given observation, we note the adaptive nature of our GNN, wherein computation is restricted to the input signal support. We demonstrate the effectiveness of our GNN architecture on a task classifying IceCube events, where it outperforms both a traditional physics-based method as well as classical 3D convolution neural networks.
KW - Deep learning
KW - Graph neural networks
KW - Pattern classification
UR - http://www.scopus.com/inward/record.url?scp=85062212454&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85062212454&partnerID=8YFLogxK
U2 - 10.1109/ICMLA.2018.00064
DO - 10.1109/ICMLA.2018.00064
M3 - Conference contribution
AN - SCOPUS:85062212454
T3 - Proceedings - 17th IEEE International Conference on Machine Learning and Applications, ICMLA 2018
SP - 386
EP - 391
BT - Proceedings - 17th IEEE International Conference on Machine Learning and Applications, ICMLA 2018
A2 - Wani, M. Arif
A2 - Sayed-Mouchaweh, Moamar
A2 - Lughofer, Edwin
A2 - Gama, Joao
A2 - Kantardzic, Mehmed
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 17th IEEE International Conference on Machine Learning and Applications, ICMLA 2018
Y2 - 17 December 2018 through 20 December 2018
ER -