SciPost Submission Page
Machine learning in top quark physics at ATLAS and CMS
by Matthias Komm
Submission summary
| Authors (as registered SciPost users): | Matthias Komm |
| Submission information | |
|---|---|
| Preprint Link: | scipost_202509_00005v1 (pdf) |
| Date accepted: | Sept. 10, 2025 |
| Date submitted: | Sept. 2, 2025, 10:06 p.m. |
| Submitted by: | Matthias Komm |
| Submitted to: | SciPost Physics Proceedings |
| Proceedings issue: | The 17th International Workshop on Top Quark Physics (TOP2024) |
| Ontological classification | |
|---|---|
| Academic field: | Physics |
| Specialties: |
|
| Approach: | Experimental |
Abstract
This note presents an overview of current and potential future applications of machine-learning-based techniques in the study of the top quark. The research community has developed a diverse set of ideas and tools, including algorithms for the efficient reconstruction of recorded collision events and innovative methods for statistical inference. Recent applications of some techniques by the ATLAS and CMS collaborations are also highlighted.
Current status:
Editorial decision:
For Journal SciPost Physics Proceedings: Publish
(status: Editorial decision fixed and (if required) accepted by authors)
Reports on this Submission
Strengths
1- The proceedings is clearly written and the presentation of ideas is easy to follow, making the review accessible to a broad audience. 2- The paper provides a well-structured and comprehensive overview of ML-based techniques that are or could be used for top physics
Report
Recommendation
Publish (meets expectations and criteria for this Journal)
Matthias Komm on 2025-09-03 [id 5776]
the following changes have been made:
Abstract, first sentence: "This note presents an overview of machine-learning-based techniques used in the study of the top quark." has been expanded to "This note presents an overview of current and potential future applications of machine-learning-based techniques in the study of the top quark."
Introduction, 2nd paragraph, first sentence: "This note reviews a collection of state-of-the-art ML algorithms addressing various aspects of top quark research, including top quark and event reconstruction, analysis strategies, and novel methods for statistical inference" has been expanded to "This note reviews a collection of state-of-the-art ML algorithms addressing various aspects of top quark research, including top quark and event reconstruction, analysis strategies, and novel methods for statistical inference, which could offer valuable contributions to the field in the future."
Both of these changes have been made to clarify that some presented techniques are not yet used in top quark research.
Top quark reconstruction, 2nd paragraph, 3rd sentence: replaced "... demonstrating ..." with "... illustrating ..." as suggested by the referee
Figure 1, caption: "Inference of the pseudorapidity of the neutrino originating from a leptonically decaying top quark" has been expanded to "Inference of the pseudorapidity of the neutrino originating from the leptonically decaying top quark in semileptonic tt̄ events" to clarify the process as requested by the referee
HL LHC outlook, 2nd paragraph, 3rd sentence: "An example variation involving the hdamp parameter of the POWHEG event generator [22] is shown in Fig. 3a. " has been expanded to "An example variation is shown in Fig. 3a involving the hdamp parameter of the POWHEG event generator [22], which regulates the energy of additional radiations. Its effect on the sample cannot be described analytically but is approximated by the NN.". This provides a more detailed explanation and in particular explains why an analytical reweighting (eg. based on matrix elements) is not viable.
HL LHC outlook, 2nd paragraph, 4th sentence: "The good agreement observed after reweighting demonstrates that this technique could replace the need for generating dedicated samples in the future." has been expanded to "The good agreement observed after reweighting demonstrates that this technique could replace the need for generating dedicated samples in the future, resulting in improved sustainability by skipping the computational needs of classical detector simulations." to underline the benefits of the approach