CERN operates some of the most complex scientific machinery ever built, relying on intricate control systems and generating petabyte upon petabyte of research data. Operating this equipment and performing analysis on the data gathered are both intensive tasks. Therefore, CERN is increasingly looking to the broad domain of artificial-intelligence (AI) research to address some of the challenges encountered in dealing with data, particle beam handling, and in the up keep of its facilities.

The modern domain of artificial intelligence came into existence around the same time that CERN did, in the mid-’50s. It has different meanings in different contexts, and CERN is mainly interested in task-oriented, so-called restricted AI, rather than general AI involving aspects such as independent problem-solving, or even artificial consciousness. Particle physicists were among the first groups to use AI techniques in their work, adopting Machine Learning (ML) as far back as 1990. Beyond ML, physicists at CERN are also interested in the use of Deep Learning to analyse the data deluge from the LHC.

As we have already mentioned, when particles collide in accelerators, numerous cascades of secondary particles are produced. The electronics that process the incoming signals from the detectors then only have a fraction of a second to assess whether an event is of sufficient interest to be stored for later analysis. In the near future, this demanding task could be taken over by algorithms based on AI.

In fact, the LHC’s four major collaborations   ALICE, ATLAS, CMS and LHCb   have formed the Inter-experimental Machine Learning (IML) Working Group to follow developing trends in ML. Researchers are also collaborating with the wider data-science community to organise workshops to train the next generation of scientists in the use of these tools, and to produce original research in Deep Learning. ROOT, the software program developed by CERN and used by physicists around the world for analysing their data, also comes with machine-learning libraries.

For exemple, although machine learning (ML) techniques were used in particle physics since the 1980’s, and in CMS since its birth, these have traditionally relied on the highly processed inputs produced by reconstruction algorithms rather than on the more basic “raw” data provided by the particle detectors. This data analysis paradigm was instrumental in several CMS physics analyses, including some interesting discoveries, but was not good enough to overcome the challenges posed by many exotic decays. In fact, the limiting factor was not so much the performance of the ML algorithms themselves but rather of the algorithms used to turn the raw detector data into basic physics observables.

Enter modern artificial intelligence (AI) and deep learning. Thanks to this newer breed of ML algorithms, it is now possible to bypass the conventional reconstruction step and directly process the detector data. Applying AI to directly extract particle information from low-level detector data (a new approach often referred to as end-to-end particle reconstruction) promises to unlock potential information pathways that could solve some of the biggest challenges we face when searching for exotic decays.

See more…

Another example is the research presented in Computer Science by scientists from the Institute of Nuclear Physics of the Polish Academy of Sciences (IFJ PAN) in Cracow (Poland) that suggests that tools built using artificial intelligence (AI) could be an effective alternative to current methods for the rapid reconstruction of particle tracks. Their debut could occur in the next two to three years, probably in the MUonE experiment at CERN which supports the search for new physics.

See more…