Erratum to: Finding new physics without learning about it: anomaly detection as a tool for searches at colliders (The European Physical Journal C, (2021), 81, 1, (27), 10.1140/epjc/s10052-020-08807-w)

Eur. Phys. J. C (2021) 81:1020 https://doi.org/10.1140/epjc/s10052-021-09813-2 Erratum Erratum to: Finding new physics without learning about it: anomaly detection as a tool for searches at colliders M. Crispim Romão1,N.F.Castro 1,2,a, R. Pedro1 1LIP, Av. Professor Gama Pinto 2, 1649-003 Lisbon, Portugal 2Departamento de Física, Escola de Ciências, Universidade do Minho, 4710-057 Braga, Portugal Received: 25 October 2021 / Accepted: 5 November 2021 / Published online: 22 November 2021 © The Author(s) 2021 Erratum to: Eur. Phys. J. C (2021) 81:27 https://doi.org/10.1140/epjc/s10052-020-08807-w On further processing the simulated data used in the article for open access release, we detected a defective feature in the benchmark signals samples. The issue affects the results of assessing the performance of the Anomaly Detection (AD) methods with the selected Beyond the Standard Model signals reported in Sect. 5— Comparison of the AD methods for benchmark signals. The correct results are presented in Figs. 1,2and Table 1. The original article can be found online at https://doi.org/10.1140/ epjc/s10052-020-08807-w. ae-mail: [email protected] (corresponding author) Although the performance of the AD methods changes quantitatively, their relative sensitivity to the benchmark signals is generally maintained: deep learning outperforms shallow learning and the Deep SVDD model presents the best response to the variety of new signals tested, indicating to better suit to the generic detection purpose. The data samples were made available for open access through the Zenodo online repository for research data, with record number 5126747 [2]. Finally, the corrected version of the article is also available in arXiv [3]. 123 1020 Page 2 of 4 Eur. Phys. J. C (2021) 81 :1020 00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 HBOS output 2− 10 1 − 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fb s pp, Full Features 00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 HBOS output 2− 10 1 − 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fb s pp, Sanitised Features 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 iForest output 2− 10 1− 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fbspp, Full Features 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 iForest output 2− 10 1 − 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fbspp, Sanitised Features 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Deep SVDD output 2 − 10 1− 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fbs pp, Full Features 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Deep SVDD output 2 − 10 1− 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fbs pp, Sanitised Features 00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 AE output 2− 10 1 − 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fbspp, Full Features 00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 AE output 2− 10 1 − 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 Events / bin SM prediction HG 1.0 TeV HG 1.2 TeV HG 1.4 TeV W/o HG 1.0 TeV W/o HG 1.2 TeV W/o HG 1.4 TeV FCNC -1 = 13 TeV, L = 150 fbspp, Sanitised Features Fig. 1 Distribution of the AD discriminant for the SM prediction and each signal type: tZ production by FCNC, T¯ Tproduction via heavy gluon or without heavy gluon for mT={1.0,1.2,1.4}TeV. The distributions are normalised to the generation cross-section and to an integrated luminosity of 150 fb−1. Left: Using all features set. Right: Using sanitised features set 123 Eur. Phys. J. C (2021) 81 :1020 Page 3 of 4 1020 Table 1 95% CL upper limit on the signal strength µof each benchmark signal for the different AD methods using the full feature set and the sanitised set and for a dedicated supervised DNN model trained on the full feature set. The statistical uncertainties, including the effect from limited statistics in the simulated datasets, are also shown Model Benchmark signal FCNC HG No HG 1.0TeV 1.2TeV 1.4TeV 1.0TeV 1.2TeV 1.4TeV Full features Supervised DNN 2.9+1.4 −0.90.09+0.04 −0.03 0.3+0.2 −0.10.17+0.07 −0.06 0.26+0.13 −0.08 1.9+1.3 −0.82.3+1.1 −0.7 HT60+20 −20 0.27+0.14 −0.09 0.3+0.2 −0.10.29+0.16 −0.09 0.8+0.5 −0.21.9+0.9 −0.73.2+1.7 −1.0 Deep SVDD 6+3 −10.4+0.1 −0.10.4+0.2 −0.10.5+0.2 −0.10.9+0.4 −0.32.6+1.2 −0.87+4 −2 AE 20+4 −90.25+0.13 −0.08 0.26+0.13 −0.08 0.28+0.13 −0.09 0.6+0.2 −0.21.4+0.7 −0.44+1 −1 HBOS 60+20 −20 0.3+0.2 −0.10.4+0.2 −0.10.4+0.2 −0.10.8+0.4 −0.22.2+1.0 −0.75+3 −1 iForest 70+30 −20 0.4+0.1 −0.20.4+0.2 −0.10.5+0.2 −0.20.9+0.4 −0.32.3+1.3 −0.76+4 −2 Sanitised features Supervised DNN 2.8+1.3 −0.90.22+0.18 −0.10.3+0.2 −0.10.4+0.2 −0.20.5+0.5 −0.21.8+1.4 −0.85+5 −2 HT50+20 −10 0.27+0.14 −0.09 0.3+0.16 −0.10.29+0.16 −0.09 0.8+0.5 −0.21.8+1.0 −0.53+2 −1 Deep SVDD 6+3 −20.19+0.08 −0.05 0.21+0.1 −0.05 0.24+0.11 −0.06 0.36+0.16 −0.09 1.1+0.5 −0.33.6+1.5 −0.9 AE 60+30 −20 0.9+0.5 −0.30.8+0.4 −0.30.6+0.4 −0.21.6+1.0 −0.54+2 −19+5 −3 HBOS 60+20 −20 0.5+0.3 −0.20.5+0.3 −0.20.5+0.3 −0.21.0+0.5 −0.42.4+1.5 −0.86+3 −2 iForest 70+30 −20 0.5+0.2 −0.20.5+0.2 −0.20.4+0.2 −0.11.1+0.5 −0.42.4+1.2 −0.85+3 −2 Fig. 2 95% CL upper limits on µnormalised to the limit obtained for the supervised DNN model 123 1020 Page 4 of 4 Eur. Phys. J. C (2021) 81 :1020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecomm ons.org/licenses/by/4.0/. Funded by SCOAP3. References 1. M. Crispim Romão, N.F. Castro, R. Pedro, Finding new physics without learning about it: anomaly detection as a tool for searches at colliders. Eur. Phys. J. C 81(1), 27 (2021) 2. M. Crispim Romao, N.F. Castro, R. Pedro, Simulated pp collisions at 13 tev with 2 leptons + 1 b-jet final state and selected benchmark beyond the standard model signals. https://zenodo.org/record/ 5126747, (2021) 3. M. Crispim Romão, N.F. Castro, R. Pedro, Finding new physics without learning about it: anomaly detection as a tool for searches at colliders (2021). arxiv:2006.05432 123