Source author record

Yoshikazu Nagai

Yoshikazu Nagai appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

hep-ex Applications Machine Learning physics.data-an physics.ins-det

Catalog footprint

What is connected

2works

5topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2020arXiv

The Forward TPC system of the NA61/SHINE experiment at CERN: a tandem TPC concept

This paper presents the Forward Time Projection Chamber (FTPC) system of the NA61/SHINE experiment at the CERN SPS accelerator. This TPC system applies a novel tandem-TPC design to reduce the background originating from particle tracks not synchronous with the event trigger. The FTPC system is composed of three chambers with alternating drift field directions. The chambers were installed directly along the beamline region of the NA61/SHINE detector in a medium- to high-intensity (10-100 kHz) hadron or ion beam. The tandem TPC system has proved to be capable of rejecting out-of-time background tracks not associated with a primary interaction. In addition, the system performs tracking and inclusive dE/dx particle identification for particles at and near the beam momentum. This shows that a tandem-TPC-based chamber design may be used also in other experimental applications with a demand for low material budget, tracking capability, and the need for dE/dx particle identification, all while in the presence of a relatively high particle flux.

preprint2012arXiv

Semi-Supervised Anomaly Detection - Towards Model-Independent Searches of New Physics

Most classification algorithms used in high energy physics fall under the category of supervised machine learning. Such methods require a training set containing both signal and background events and are prone to classification errors should this training data be systematically inaccurate for example due to the assumed MC model. To complement such model-dependent searches, we propose an algorithm based on semi-supervised anomaly detection techniques, which does not require a MC training sample for the signal data. We first model the background using a multivariate Gaussian mixture model. We then search for deviations from this model by fitting to the observations a mixture of the background model and a number of additional Gaussians. This allows us to perform pattern recognition of any anomalous excess over the background. We show by a comparison to neural network classifiers that such an approach is a lot more robust against misspecification of the signal MC than supervised classification. In cases where there is an unexpected signal, a neural network might fail to correctly identify it, while anomaly detection does not suffer from such a limitation. On the other hand, when there are no systematic errors in the training data, both methods perform comparably.