Deliverable 4.2 Is Out

by Tommaso Dorigo

I am happy to report that an important new product of the AMVA4NewPhysics ITN is now public. This is generically titled “Report on a Statistical Learning Method for Model-Independent Searches for New Physics“, and labeled D4.2 as per the grant agreement we signed with the European Union. The document is available at the following link:

http://www.pd.infn.it/%7Edorigo/D4.2.pdf

What is this document about ? It is a description of the studies for the development of a software package aiming at automating the searches for new physics in LHC data, by evidencing anomalous clusterings of events that are hard to explain with known physics processes. I am sure that Fabricio and Grzegorz, the two main developers of the software (Deliverable 4.3, available on github at https://github.com/Grzes91/PenalizedAD) and its documentation (D4.2) will be happy to post in this blog a more complete description of the new package and its possible uses in particle physics research.

Convolutional Neural Networks and neutrinos

by Cecilia Tosciri

Have you ever wondered how Facebook suggests the tags for the picture you post on your wall, or how the photo library on your computer manages to automatically create albums containing pictures of particular people? Well, they use facial recognition software based on Convolutional Neural Network (CNN).

CNN is the most popular and effective method for object recognition, and it is a specialized kind of neural network for processing data that has a known grid-like topology. The network employs a mathematical operation

Analysis of Italian Roads

by Greg Kotkowski

It is said that “all roads lead to Rome”. Is it true anymore? Certainly, during the Roman Empire main roads were constructed in such the way that everybody could easily reach the capital, the political and economical center of the country. Therefore if roads are built in order to facilitate the transportation toward the most important hubs, they could be used as an indicator of a region’s importance.

I downloaded the data of all contemporary roads in Italy from the OSM. As a starter, it is worth to plot them all (see Figure 1). It is

Do Not Name Him Donald!

by Grzegorz Kotkowski

Recently I’ve encountered an interesting article about the trends of the female names in the US. It shows the impact of the famous Disney Movies on the names that are given to the newborns. As the “Frozen” movie has become very popular a lot of girls born in 2014 got names as Elsa or  Merida.

I want to consider the same dataset in order to perform the analogous analysis but for names of the US presidents. My guess is that it should well represent if a

Smoothing alien data

by Pablo de Castro

What if you have some data you want to model, but do not know anything about its parent distribution, so you have to make as little assumptions as possible? In this post, I will go through the concept of density estimation and I will play with some interesting non-parametric methods.

Secondment in Padua

by Cecilia Tosciri

A few days ago I left Padua, where I spent one intensive month, working with other network members and ESR fellows (Giles, Greg and Pablo) at the Statistical Department.

A first observation

When people from different disciplines work together, like in the case of physicists and statisticians, the first stumbling block is the communication. This is mostly because

Mixture of normals

by Greg Kotkowski

Modern statistical modelling seeks for more and more flexible methods to describe a wide variety of random phenomena. The Gaussian distribution is heavily exploited thanks to its properties, easy interpretation and simplicity. However, the data is often more complex and fitting it with a normal distribution is insufficient for skewed or heavily-tailed settings. Hence, more sophisticated methods are of great importance.

On the other hand, more complex approaches bring new difficulties. It is often not as straightforward