Machine learning has proven to be an indispensable tool in the selection of interesting events in high energy physics. Such technologies will become increasingly important as detector upgrades are introduced and data rates increase by orders of magnitude.
HEPDrone is a toolkit to enable the creation of a drone classifier from any machine learning classifier, such that different classifiers may be standardised into a single form and executed in parallel. This involves the creation of a drone neural network, which learns the required properties of the input neural network without the use of any training data, only using appropriate questioning of the input neural network.
While this technology has been proven to work converting one deep neural network to another the full extent of the drone suitability has yet to be determined. Work will need to be undertaken to give more flexibility to the drone model, i.e. giving more options for drone network types such as convolutional neural networks and long short-term memory capability. A detailed evaluation of the performance of different drone models in the real production environment of LHCb will give the collaboration a complete idea of not only the advantages of the drone model, but also the limits of drone complexity given the available computing resources.