BEGIN:VCALENDAR VERSION:2.0 PRODID:-//linuxsoftware.nz//NONSGML Joyous v1.4//EN BEGIN:VEVENT SUMMARY:The impact of Artificial Intelligence on precision Higgs Boson phy sics DTSTART:20250326T160000Z DTEND:20250326T180000Z DTSTAMP:20260905T135444Z UID:0045ba8d-c640-4239-9727-89d8b306a801 SEQUENCE:1 CREATED:20250318T145014Z DESCRIPTION: Particle physics has a long history of developing sophisticat ed simulators\, from modeling proton collisions to creating virtual detect ors. Now\, we'\;re integrating Artificial Intelligence through "\; Neural Simulation Based Inference\,"\; which helps analyze complex phe nomena that cannot be to computed analytically from first principles. Thi s approach is highlighted by recent work from the ATLAS experiment at the Large Hadron Collider at CERN\, which has provided new insights into quant um interference in a rare Higgs boson decay. We also touch on how biases ( "\;known unknowns"\;) are modeled\, as showcased in the recent Fai r Universe HiggsML uncertainty challenge at NeurIPS 2024. Separately\, ad vances in AI generative models\, which allow to generate fake images\, can be harnessed to emulate particle physics simulators\, albeit with specifi c challenges which will be presented. LAST-MODIFIED:20250318T145014Z LOCATION:Anfiteatro PA1 (Piso -1 do Pavilhão de Matemática) do IST URL:http://df.vps.tecnico.ulisboa.pt/en/events/the-impact-of-artificial-in telligence-on-precision-higgs-boson-physics/ X-ALT-DESC;FMTTYPE=text/html:

Particle physics h as a long history of developing sophisticated simulators\, from modeling p roton collisions to creating virtual detectors. Now\, we'\;re integrat ing Artificial Intelligence through "\;Neural Simulation Based Inferen ce\,"\; which helps analyze complex phenomena that cannot be to comput ed analytically from first principles.

 This approach is highlig hted by recent work from the ATLAS experiment at the Large Hadron Collider at CERN\, which has provided new insights into quantum interference in a rare Higgs boson decay. We also touch on how biases ("\;known unknowns "\;) are modeled\, as showcased in the recent Fair Universe HiggsML un certainty challenge at NeurIPS 2024.

 Separately\, advances in A I generative models\, which allow to generate fake images\, can be harness ed to emulate particle physics simulators\, albeit with specific challenge s which will be presented.

END:VEVENT END:VCALENDAR