Speaker
Description
In 2012 --the same year when ML methods proven super-human image classification power in the ImageNet challenge-- the CMS and ATLAS collaborations employed for the first time supervised learning tools for a major physics discovery (the Higgs boson). That constituted a revolution in how inference is extracted from complex data in high-energy physics: without ML tools before 2012, with ML tools after it.
A new revolution is now about to start, as artificial intelligence (AI) today allows machines to help us carry out the end-to-end, goal-oriented optimization of our experiments, achieving full co-design of the geometry and specifications of hardware instruments producing the raw data together with the details of software algorithms performing pattern recognition, dimensionality reduction, and inference. In this talk we will examine the status of this impending paradigm change, and the technical hurdles that need to be overcome to realize it.
Details
Dr. Tommaso Dorigo, INFN - Sezione di Padova, Italy
https://userswww.pd.infn.it/%7Edorigo
| Internet talk | No |
|---|---|
| Is this an abstract from experimental collaboration? | No |
| Name of experiment and experimental site | N/A |
| Is the speaker for that presentation defined? | Yes |