Conveners
Machine Learning & AI: New Physics
- Wen Han Chiu (UIUC)
Machine Learning & AI: Collider Physics
- Prasanth Shyamsundar (Fermi National Accelerator Laboratory)
-
Tae Min Hong (University of Pittsburgh (US))13/05/2024, 14:00Machine Learning & AI
We present a decision tree-based implementation of autoencoder anomaly detection. A novel algorithm is presented in which a forest of decision trees is trained only on background and used as an anomaly detector. The fwX platform is used to deploy the trained autoencoder on FPGAs within the latency and resource constraints demanded by level 1 trigger systems. Results are presented with two...
Go to contribution page -
Chosila Sutantawibul (Baylor University (US))13/05/2024, 14:15Machine Learning & AI
The Compact Muon Solenoid (CMS) detector at the CERN LHC produces a large quantity of data that requires rapid and in-depth quality monitoring to ensure its validity for use in physics analysis. These assessments are often done by visual inspection which can be time consuming and prone to human error. In this talk, we introduce the “AutoDQM” system for Automated Data Quality Monitoring in CMS...
Go to contribution page -
Ranit Das (Rutgers University)13/05/2024, 14:30Machine Learning & AI
We present R-Anode, a new method for data-driven, model-agnostic resonant anomaly detection that raises the bar for both performance and interpretability. The key to R-Anode is to enhance the inductive bias of the anomaly detection task by fitting a normalizing flow directly to the small and unknown signal component, while holding fixed a background model (also a normalizing flow) learned from...
Go to contribution page -
Hancheng Li (University Of California, Santa Barbara)13/05/2024, 14:45Machine Learning & AI
Anomaly detection is a promising, model-agnostic strategy to find physics beyond the Standard Model. State-of-the-art machine learning methods offer impressive performance on anomaly detection tasks, but interpretability, resource, and memory concerns motivate considering a wide range of alternatives. We explore using the 2-Wasserstein distance from optimal transport theory, both as an anomaly...
Go to contribution page -
Radha Mastandrea (LBNL)13/05/2024, 15:00Machine Learning & AI
Understanding the higgs boson, both in the context of Standard Model physics and beyond-the-Standard Model hypotheses, is a key problem in modern particle physics. An increased understanding could come from the detection and analysis of pairs of higgs bosons produced at hadron colliders. While such higgs pairs have not yet been observed at the Large Hadron Collider (LHC), it is likely that...
Go to contribution page -
Umar Sohail Qureshi (Vanderbilt University)13/05/2024, 15:15Machine Learning & AI
A model based on a $U(1)_{T 3R}$ extension of the Standard Model can address the mass hierarchy between the third and the first two generations of fermions, explain thermal dark matter abundance, and the muon $g - 2$ and $R_{K^{(*)}}$ anomalies. The model contains a light scalar boson $\phi'$ and a heavy vector-like quark $\chi_u$ that can be probed at CERN's Large Hadron Collider (LHC). We...
Go to contribution page -
Wen Han Chiu (UIUC)14/05/2024, 14:00Machine Learning & AI
We present work in progress on using the timing information of jet constituents to determine the production vertex of highly displaced jets formed from the decay of a long-lived particle. We also demonstrate that the same network can output a much more consistent jet time that is less sensitive to geometric effects; allowing for better exclusionary power compared to $p_T$-weighted time.
Go to contribution page -
Ahmed Youssef (University of Cincinnati)14/05/2024, 14:15Machine Learning & AI
Hadronization, a crucial component of event generation, is traditionally simulated using finely-tuned empirical models. While current phenomenological models have achieved significant success in simulating this process, there remain areas where they fall short of accurately describing the underlying physics. An alternative approach is machine learning-based models.
Go to contribution page
In this talk, I will... -
Austin Edwin Townsend (University of Notre Dame (US))14/05/2024, 14:30Machine Learning & AI
New physics at the LHC may be hiding in non-standard final state configurations, particularly in cases where stringent particle identification could obscure the signal. Here we present a search for resonances in the three-photon final state where two photons are highly merged. We target the case where a heavy vector-like particle decays to a photon and a new spin-0 particle $\phi$, where...
Go to contribution page -
Ian Pang14/05/2024, 14:45Machine Learning & AI
Normalizing flows have proven to be state-of-the-art for fast calorimeter simulation. With access to the likelihood, these flow-based fast calorimeter surrogate models can be used for other tasks such as unsupervised anomaly detection and incident energy regression without any additional training costs.
Go to contribution page -
Kyle Angelo Granados (California State University (US))14/05/2024, 15:00Machine Learning & AI
The invariant mass of particle resonances is a key analysis variable for LHC physics. For analyses with di-tau final states, the direct calculation of the invariant mass is impossible because tau decays always include neutrinos, which escape detection in LHC detectors. The Missing Mass Calculator (MMC) is an algorithm used by the ATLAS Experiment to calculate the invariant mass of resonances...
Go to contribution page -
Kyungmin Park (Carnegie-Mellon University (US))14/05/2024, 15:15Machine Learning & AI
In a search for an exotic Higgs boson decay, a novel signature with highly collimated photons is studied where the Higgs boson decays into hypothetical light pseudoscalar particles of the form H to AA. In the highly boosted scenario, two collimated photons from the A decay are reconstructed as a single photon object, or an artificially merged photon shower. A deep learning based tagger is...
Go to contribution page