| Description: | Upcoming Fast Radio Burst surveys will search thousands of beams at 10^4 DMs, nearly 24/7. Searching such a large phase space results in an enormous number of pulse candidates. The production of such false positives must be mitigated, and their classification must be automated. In a recently-published article we describe a new set of tools using deep learning to classify in real-time Fast Radio Burst candidates (Connor & van Leeuwen 2018). This figure shows the tree-like architecture of our multi-input deep convolutional neural network. |
