Ref [1] presented a method to construct novel observables exclusively from information identified by the machine, by using a parametrization of the N-body phase space coordinates at the point of saturation of discrimination power. We have now studied how to extend this approach in an automated way to higher N-body phase space. We use boosted Z' vs QCD discrimination as a testing ground...
In the current era of high energy experiments, we are faced with an overwhelming amount of data and the limiting uncertainty in new physics searches can often come from theory and not experiment. In our efforts to develop new approaches to extract complex signals from large backgrounds, BDTs, neural networks and other machine learning techniques are becoming increasingly significant. These...
Machine learning in high energy physics relies heavily on simulation for fully supervised training. This often results in sub-optimal classification when ultimately applied to (unlabeled) data. In addition to describing a new method for weak supervision (learning directly from data) called Classification Without Labels (CWoLa), we show for the first time how to apply these techniques to...
Classification Without Labels (CWoLa) is a Machine Learning strategy which can be used to classify event categories (e.g. quark jet vs gluon jet, or BSM signal vs SM background) starting from mixed event samples, which are inevitable at particle collider experiments. I will illustrate how this strategy can be used to uncover BSM resonance signals which would otherwise be completely buried...
We would like to present a study on the behavior of Recursive Neural Networks (RecNNs) in jet tagging. The RecNNs embed jet clustering history recursively to include all the information within a jet. We examine its behavior in different jet tagging tasks, and analyze the connection between RecNNs and the underlying physics.
In applications of machine learning to particle physics, there is a persistent tension between interpretability and performance. In this paper, this tension is allayed by introducing a novel framework for unsupervised machine learning in particle physics, in which the neural network architecture is built as a scaffolding around a leading-order description of the physics under study. This...
Recent results at the Large Hadron Collider (LHC) have pointed to enhanced physics capabilities through the improvement of the real-time event processing techniques. Machine learning methods are ubiquitous and have proven to be very powerful in LHC physics, and particle physics as a whole. However, exploration of the use of such techniques in low-latency, low-power FPGA hardware has only just...
