BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:UW-Madison-Physics-Events
BEGIN:VEVENT
SEQUENCE:0
UID:UW-Physics-Event-8823
DTSTART:20240730T190000Z
DTEND:20240730T210000Z
DTSTAMP:20260413T190948Z
LAST-MODIFIED:20240712T013719Z
LOCATION:5310 Chamberlin Hall
SUMMARY:Search for exotic Higgs boson decays with CMS and fast machine
  learning solutions for the LHC\, Thesis Defense\, Ho Fung Tsoi\, Phys
 ics PhD Graduate Student
DESCRIPTION:The first part of this thesis presents a search for new ph
 ysics with the CMS experiment. There is still potential for discoverie
 s beyond the Standard Model in the scalar sector\, which could manifes
 t as exotic Higgs boson decays into light pseudoscalars. This search t
 argets such decays\, focusing on pseudoscalar masses ranging from 12 a
 nd 60 GeV\, in final states where one pseudoscalar decays into two b q
 uarks and the other into two $\\tau$ leptons or two muons. The analysi
 s is based on a dataset of proton-proton collisions at $\\sqrt{s}=13$ 
 TeV\, collected by the CMS detector during LHC Run 2\, with an integra
 ted luminosity of 138 $\\text{fb}^{-1}$. Dedicated neural networks are
  used to distinguish between signal and background\, significantly enh
 ancing sensitivity. The results are presented as exclusion limits at 9
 5\\% confidence level on the model-independent branching ratio and are
  interpreted within two-Higgs doublet models augmented by a singlet. T
 he second part of this thesis presents machine learning methods to enh
 ance overall sensitivity in the low-latency domain for the LHC experim
 ents. A novel machine learning-based trigger algorithm is developed\, 
 using anomaly detection to search for new physics in a model-agnostic 
 manner as close to the raw collision data as possible. This anomaly de
 tection trigger is sensitive to a wide range of both conventional and 
 unconventional physics signals and has an inference latency of O(100) 
 ns on an FPGA. It is deployed during Run 3 in the CMS Level-1 trigger 
 system\, which processes the first round of real-time event selection 
 from collision data at a rate of 40 MHz. Additionally\, a novel model 
 compression method using symbolic regression is developed to accelerat
 e machine learning inference to nanosecond speeds on FPGAs. We demonst
 rate its potential to significantly reduce the computational costs of 
 machine learning algorithms while maintaining performance comparable t
 o that of neural networks. These advancements are crucial for meeting 
 the sensitivity and computational demands of resource-constrained envi
 ronments such as the LHC experiments.
URL:https://www.physics.wisc.edu/events/?id=8823
END:VEVENT
END:VCALENDAR