Cargando…
Machine learning techniques for razor triggers
My project was focused on the development of a neural network which can predict if an event passes or not a razor trigger. Using synthetic data containing jets and missing transverse energy we built and trained a razor network by supervised learning. We accomplished a ∼ 91% agreement between the out...
Autor principal: | |
---|---|
Lenguaje: | eng |
Publicado: |
2015
|
Materias: | |
Acceso en línea: | http://cds.cern.ch/record/2046071 |
Sumario: | My project was focused on the development of a neural network which can predict if an event passes or not a razor trigger. Using synthetic data containing jets and missing transverse energy we built and trained a razor network by supervised learning. We accomplished a ∼ 91% agreement between the output of the neural network and the target while the other 10% was due to the noise of the neural network. We could apply such networks during the L1 trigger using neuromorhic hardware. Neuromorphic chips are electronic systems that function in a way similar to an actual brain, they are faster than GPUs or CPUs, but they can only be used with spiking neural networks. |
---|