A neural network for recognizing Cerenkov radiation patterns in space and time
Journal Article
·
· Bulletin of the American Physical Society
OSTI ID:387224
- Lawrence Berkeley Laboratory, CA (United States)
We have studied the feasibility of using a neural network to characterize events obtained from large Cerenkov imaging detectors (such as SNO). A single event is comprised of a certain number of {open_quote}fired{close_quote} phototubes distributed both spatially and temporally. The tasks for the network are (i) to discriminate real events from noise, and (ii) to separate events with vertices inside or outside a region of interest. A computational three layer feed-forward network made up of one input layer, two hidden layers, and an output layer has been devised and tested for this purpose. The first layer connections are derived from the method of principal component analysis (PCA) and the connections of the second and third layers are trained by back error propagation (BEP). The performance of the network compared to a conventional chi-square fitting algorithm for low energy Monte-Carlo events will be discussed.
- DOE Contract Number:
- AC03-76SF00098
- OSTI ID:
- 387224
- Report Number(s):
- CONF-931044--
- Journal Information:
- Bulletin of the American Physical Society, Journal Name: Bulletin of the American Physical Society Journal Issue: 9 Vol. 38; ISSN BAPSA6; ISSN 0003-0503
- Country of Publication:
- United States
- Language:
- English
Similar Records
Do neural networks offer something for you?
Single-Iteration Learning Algorithm for Feed-Forward Neural Networks
Wide neural networks with bottlenecks are deep Gaussian processes
Journal Article
·
Tue Oct 31 23:00:00 EST 1995
· InTech
·
OSTI ID:443471
Single-Iteration Learning Algorithm for Feed-Forward Neural Networks
Conference
·
Sat Jul 31 00:00:00 EDT 1999
·
OSTI ID:6257
Wide neural networks with bottlenecks are deep Gaussian processes
Journal Article
·
Mon Aug 31 20:00:00 EDT 2020
· Journal of Machine Learning Research
·
OSTI ID:1820877