by Tommaso Dorigo

The AMVA4NewPhysics network has reached month 17, and its third solar year in existence. It is time, I think, to make a short summary of what has been going on until now, before we concentrate our attention on what lies ahead.

Overall, if we look at the plan as its authors designed it in the fall of 2014, we see an impressive level of agreement. This is by no means a trivial statement! Writing a successful ITN project involves making a number of predictions about the future, and imagining the development of research and training beyond what is usually possible. Hence the coincidence of plan and actual action is to be remarked. Let me list below some of the things that went perfectly along according to the plan, and also things that required some steering, to summarize what we have left behind our back.

– We had three well-attended and fruitful supervisory board meetings in close temporal match with the schedule. The first was held at CERN in September 2015, the second in Venice in March 2016, and the third last December in Oxford. While the first meeting was mainly organizational and focus on warming up our muscles, the second focused on maximizing the network members’ participation and creating a public engagement opportunity (which was absolutely a success, with Daniela Bortoletto’s lecture with music and refreshments to an audience of 150 laypersons!). That event also allowed us to get acquainted with the Project Officer of the REA (Research Executive Agency of the EU), our supervisor and counselor. As for the third meeting in Oxford, we started to shift our focus on research output, with many interesting scientific discussions taking place in the perfect atmosphere of a well-functioning network.

– We are proceeding along the plan with our outreach action: this blog has established itself as a place where outsiders can read what is the life of an early-stage researcher doing his or her PhD in particle physics or statistics in a EU network, and learn some physics and machine learning in the meantime. Our press office coordinator Sabine Hemmer has been instrumental in making this possible. Of course, keeping the blog lively requires a constant effort from all the parties involved – something I never tire to explain to our students and network members!

– Speaking of outreach, I should also mention the twitter account @AMVA4NP and its hashtag #AMVA4NP. Thanks to our outreach officer Pietro Vischia the tweeting action continues steadily, although I believe we should aim for a different setup whereby everybody can contribute to it more directly.

– Outreach goes along well with Dissemination. Here I can confidently say that we over-delivered, as we decided to invest network funds to sponsor a parallel session at a physics conference in Thessaloniki last September, something we had not foreseen in the proposal (and something which could not be promised in good faith there!). Indeed, hijacking a physics conference to add a parallel session titled “Statistical methods for physics analysis in the XXI century” was a rather unconventional and hard-to-implement move. We were however quite successful, as the session sported 17 excellent presentations by leading experts on the field, and attracted a significant interest. In fact there are thoughts of repeating that feat in the near future.

– A number of other products have appeared in 2016 thanks to the network action. I should certainly mention the important MoMeMta software package, to which ESR Alessia Saggio and the Louvain group contributed significantly, the preliminary result of a CMS search for HH->bbbb decays to which ESR Pablo de Castro gave a crucial input with INFN-PD, a similar result by Oxford for ATLAS, the publication of a proceedings paper on a new ML tool for anomaly detection by Pietro Vischia from Oviedo, the talk by University of Padova’s Bruno Scarpa in Thessaloniki on Bayesian non-parametric classification methods, Pablo’s similar talk on a new background-shape estimation method in the same venue, the issuing of an internal CMS note on the same topic, the production of a innovative flavor-tagging algorithm by the CERN group with the contribution of ESR Anna Stakia… And I could go on.

– Secondments in partner companies have started as scheduled (although a couple have been delayed for technical reasons): ESR Pablo de Castro visited SDG in Milano, where he developed visualization tools for topological multiscale analysis; and ESR Alessia Saggio spent three months in B12 working at ML applications for specific problems of that company. Other inter-beneficiary secondments have also taken place exactly as scheduled: several ESRs met in Oxford and Padova to produce simulated samples of data for the study of classification and regression algorithms, and ESR Grzegorz Kotkowski visited CERN for two months, to learn about physics (he is the only statistician ESR in the network) and to study the statistical properties of one important algorithm we are about to deliver.

– Recruitment has followed the plan, but here the road has been bumpier than anticipated, for a couple of reasons. In IASA there has been a delay due to the change of PI of the node (many thanks to Niki Saoulidou for her outstanding work as equal opportunities officer as former PI in IASA, and welcome to Constantinos Vellidis as new PI!), and the ESR there is going to be hired this month, with a delay of a few months which will however not impact the network plan. And TUM (Technische Universitat Munchen) has become a new beneficiary because of the new position as associate Professor that was achieved by Andreas Weiler there (congratulations Andreas!); the change caused a delay of hiring of the ESR who will work with Andreas, who has been selected and will start working for AMVA4NewPhysics in a few weeks.

– As far as events go, we were very successful. We organized a Machine Learning and MATLAB course at CERN which was open to everybody free of charge, and was taught by a true expert in the field, MathWorks’ Ilya Narsky, a former particle physicist. The event attracted a very large participation (in fact we had to find a larger conference room than anticipated!) and was very well received. We also sponsored a Statistics School by IN2P3 in Autrans, where network members Donini and Dorigo gave lectures. As foreseen in our plan we benefited from the Erice School of Science Journalism by having our press office coordinator Sabine attend it (she won a grant to attend the event so it did not even cause the network to pay for her participation!). We sent our ESRs to the Yandex school of data analysis in Lund as foreseen. One ESR attended a conference in Crete (ICNFP) in July, and gave a talk in Thessaloniki in another (I know I said that already, but it is important to stress this!). Louvain organized workshops in matrix-element methods. One member of INFN-PD, Martino dall’Osso, was selected to show a network product (the analysis of HH->bbbb) at a conference in China. Oxford wrote a paper on the use of neural networks for the selection of higgs pair production. And I am surely forgetting things here… A lot has happened in 2016, really!

With 2016 behind our backs, we can surely be confident on our means for this year. The first thing to worry about is the successful delivery of three important deliverables: the organization of a Soft Skills workshop, which will take place in Padova on February 1 to 3, and will see the participation of most of our ESRs plus seven PhD students from the department of Physics and Astronomy in Padova. And two documents describing classification and regression tools for the Higgs search, and background modeling methods using MVA. Work on these deliverables is intensely going on, but we need to converge now!

Farther along, I should mention the secondment of Pablo and Anna at the Center for Machine Learning in UCI, where they will benefit from the great knowledge of our partner Daniel Whiteson.

Another thing coming soon, which is going to turn into a new success and a further proof of over-delivery of our outstanding ITN network, is the participation through a sponsorship to the international EPS conference, which will take place in the Lido of Venice in July 2017. Being a member of the local organizing committee of that conference I was able to offer the contribution of our network to help organize the involvement of high-school students in Venice, by offering lectures by our ESRs and a competition for the production of artwork by the students, inspired by the physics gadgets and research we do at the LHC. The plan is the following:

  1. ESRs will give talks to high-school students participating in the initiative at schools in Venice (the Liceo Foscarini, the Benedetti-Tommaseo, the Marco Polo, the Stefanini, and the Bruno-Franchetti) at the end of January. These will include showing inspiring visuals (event display, detector pictures)
  2. Students will form small groups and together they will spend several hours in the months of February, March and April to produce their artwork. The hours spent in this project will count for the “Alternanza Scuola Lavoro” that Italian students have to attend to in the last years of high school.
  3. a committee formed by a member of the CREATIONS EU network, Dorigo, and the professor of Contemporary Art in Padova Ivan Bianchi, plus local professor from the schools, will select the best artwork in each school.
  4. The selected works will be exhibited in July at the venue of the EPS conference. There, a final selection will identify the recipients of prizes – consumer electronics gadgetry (offered by the network with its sponsorship).

As you see, this is an innovative way to do public engagement, and it kills several birds with the same stone: we do outreach, we collaborate to make EPS 2017 a success, we allow our ESRs to train themselves to speak in public, and we deliver one product we promised to the EU – in fact the lectures by ESRs in high schools is a deliverable of our network!

Besides the above, much more is in store… So stay tuned for more chapters of our challenging adventure!