Past ATLAS Support Center Events
This Machine Learning and b-tagging workshop targets improvements in b-jet identification, including its basic impact parameter and vertex reconstruction algorithms, which can be gained from ML tools such as deep learning, sequence learning and domain adaptation techniques. Within ATLAS, modern ML techniques have begun to be studied for b-tagging but have yet to be fully understood and exploited.
The workshop will introduce modern ML and b-jet identification algorithms to participants through talks and tutorials, while most of the workshop time will be dedicated to working on projects proposed by participants or meeting organizers. The may include the use of Recurrent Neural Networks and sequence learning algorithms for impact-parameter or secondary/tertiary vertexing algorithms, integration of new ML techniques into high-level b-taggin algorithms like MV2 and DL1, mitigating systematics of the high-level tagger at training time, or the development of new double b-tagging algorithms.
This meeting is focused on using Geant4 to understand the response of MicroMega chambers to low-energy neutrons and photons, which are the dominant background in the New Small Wheel. Interested participants should have some familiarity with standalone Geant4. They should apply for a SLAC computer account before arrival.
The 2016 US ATLAS Hadronic Final State Forum (HFSF2016) takes place at SLAC National Accelerator Laboratory. The workshop is open to all (US) ATLAS collaborators. The workshop organization follows previous meetings of this series. More program details will be announced as they become available.