iCSC2008 Special Topics

Details of all lectures

Transition between HPC and Data Analysis themes: Using HPC concepts in Data Analysis software

   

Tuesday 4 March

 

15:10-16:25

HPC
Theme closing

Using HPC concepts in data analysis software

Alfio Lazzaro

A short session (15 minutes) to connect to the  data analysis software and techniques lecture

Some possible applications of parallel processing in data analysis code are briefly presented (e.g. how to speed up Maximum Likelihood fits).

Audience and Pre-requisite
Attendees are expected to have
some experience in data analysis.Having followed part of the HPC theme will help

Overview of advanced aspects of data analysis software and techniques

   

Wednesday 5 march

 
09:00 - 09:55 Lecture 1

Overview of advanced aspects of data analysis software and techniques

 

Alfio Lazzaro

Summary
In this lecture we give an overview of the advanced data analysis techniques based on multivariate techniques, which are recently used in many High Energy Physics data analysis. The topic is relevant to many Particle Physics analysis, as well as in several other fields.  We will give an overview on the different techniques and their relative merits.

Audience
This lecture targets an audience with experience in data analysis, in particular interested in techniques of signal/background discrimination

Pre-requisite
This lecture can be reasonably followed without having attended to the other lecturers of this school

Keywords

  • Data analysis

  • Parallel processing

  • Signal Background Separation

  • Maximum Likelihood

  • Artificial Neural Network

  • Decision Tree

Details

In the past years, many advanced techniques in statistical data analysis have been used in High Energy Physics (such as maximum likelihood fits, Neural Networks, and Decision Trees). In the past, the most common technique was the simple cut and count analysis. This technique consists in the following steps: several cuts are applied on well studied discriminating variables, background estimation is performed using Monte Carlo simulation samples or events outside the signal region, and then the final measurement is done counting the events after cuts minus the estimated background events.

 

This simple technique is hampered by its  low efficiency (defined as ratio between the number events after and before the cuts) and does not provide a good discrimination between signal and background events. For this reason it was replaced by more sophisticated techniques, such as the multivariate maximum likelihood for the measurements done at the BaBar experiment, running at Stanford Linear Accelerator Center (SLAC) in California.

 

The maximum likelihood (ML) technique permits to achieve higher efficiency, the possibility to take in account errors with a better precisions, and consider correlations between the discriminating variables used in the analysis. Anyway, in future experiments, like LHC experiments at CERN, it may be crucial to have better discrimination between signal and background events to discover new phenomenas, which suffer higher background. Neural Networks and Decision Trees are good techniques to reach this goal. Another important issue to take into account lies in the fact  that these techniques are in most cases very CPU-time consuming. It is possible to speed them up using concepts of High Performance Computing (HPC).

 

In this lecture we will give an overview of the advanced data analysis techniques mentioned above, introducing some software packages commonly used in HEP. This will be preceded by a short session at the end of the previous theme,  giving briefly examples of possible HPC optimizations.

Scalable Image and Video coding

   

Wednesday 5  march

 

10:30 - 12:00

Lecture2

Scalable Image and Video coding

Jose Dana Perez

The aim of this lecture is to describe the basis of image and video coding and compression, with a special emphasis on the latest developments. We will see how to encode and compress this particular type of  data using lossy algorithms that take advantage of the limitations of the human visual system.

We will focus on scalable image and video coding, which is a cutting-edge area of research, an area were few fully recognized standards have emerged yet. 

Sometimes, specialized  developers need to design systems which require an image or video (de)coder. Understanding the internals of some coding systems may help them in  to select the most appropriate approach (streaming systems, pattern recognition systems, etc.) and  algorithm (JPEG, JPEG2000, MPEG-2, MPEG-4, WMV, etc.).

We will present techniques used in well-known algorithms and the audience will have the opportunity to learn the fundamentals through practical examples.

Audience
TBW

Pre-requisite
TBW