MACHINE LEARNING AND AI

The integration of automatic learning or machine learning – i.e. the capacity of a machineto learn and improve its performance with experience – in modelling physical systems and analysing experimental data is offering very interesting opportunities (in some cases, revolutionary ones) for the scientific community. One of the most significant applications concerns the processing of data collected in large experiments, like particle physics ones. Machine learning algorithms are now used habitually to identify significant signals from data, significantly accelerating the analysis process compared to conventional approaches.

And this isn’t all. The identification of particles colliding at very high energy can also be improved through recognition algorithms based on machine learning, which is more efficient compared to conventional methods. In addition, automatic learning is now also exploited in implementing complex physical models, enabling faster and more detailed simulations.
More generally, artificial intelligence (AI, of which machine learning can be considered a sub-set) is experiencing an unprecedented period of progress, which promises to revolutionise physics research too at an even more profound level. One AI frontier could even be the capacity to discover new physics laws or direct new lines of research, thanks to the possibility of identifying correlations that are not clear, both in analysing data and computational simulation of physical systems.
As well as being involved in many research projects that develop and use AI and machine learning systems in the field of physics, INFN has recently launched the national initiative ML_INFN. This aims to improve the expertise of researchers in using technologies based on artificial intelligence, providing a common hardware platform and the possibility of sharing already existing knowledge at various levels within the community.

Picture generated with AI

You might also be interested in

Quantum computers

Quantum computers have always been considered a kind of “holy grail” of technological innovation, capable of revolutionising the IT world and all of society thanks to their applications.

Computing and supercomputing

Today, computers and supercomputers are essential tools for physics research, with the purpose of analysing experimental data and implementing theoretical models useful for studying often very complex phenomena, impossible to investigate with conventional analytical tools.

Physics and big data

The increase in the sensitivity and efficiency of physics experiments, combined with the use of more and more technologically advanced electronics, has led to an explosion in the quantity of data collected during experiments in recent decades.

Computing and networks

For many years, INFN has developed its own infrastructure dedicated to scientific computing. Both analysing data produced by the big experiments and theoretical simulations actually need computing power, large storage quantities, and ultra-fast networks.

Models and simulations

Since the dawn of the scientific investigation of natural phenomena, the construction ofsimplified models of physical reality has been an essential tool for studying more or less complex systems.