31641 - Analysis of Physical Data

Learning outcomes

To learn and to apply statistical methods for data analysis  and pattern recognition methods.

Course contents

Probability,Random Variables,Probability density Function,Cumulative Function,Bayes Formula,moments of a distribution,statistical tests,parametric correlation,non parametric correlation,modeling of data,introduction to pattern recognition,features reduction and features extraction,genetic algorithms,simulated annealing,training and test,k fold cross validation,Parzen windows,KNN,neural networks,peceptron,multilayer perceptron,RBF neural networks,Support Vector Machine,Cluster analysis.Deep learning.

Readings/Bibliography

Numerical recipes.The art of scientific computing,3rd edition (chapters 14 and 15) www.nr.com

B.R.Martin,"Statistics for Physical Sciences,An introduction",Elsevier,2012

G.Bohm,G.Zech " Introduction to statistics and data analysis for physicist" www-library.desy.de/preparch/books/vstamp_engl.pdf

Statistics course Home Page,Glenn Cowan:

http://www.pp.rhul.ac.uk/~cowan/stat_course.html

prof. Ricardo Gutierrez Osuna Lectures on Pattern Recognition

http://research.cs.tamu.edu/prism/lectures.htm

"Learning from data,a short course",Yaser S.Abu-Mostafa et al, AMLbook.com,2012

Free Online book on deep neural networks at http://www.deeplearningbook.org

Teaching methods

Class lectures.

Assessment methods

The student can choice two written midterm assessments or a final written total  test. At the end the student will present a computer program on course topics or related to course topics.

Teaching tools

First semester period,Dipartimento di Fisica,Viale B.Pichat 6/2

Technical support material for the computer program is available at:

Google site uniboimage http://sites.google.com/site/uniboimage/

Good online course on Convolutional Neural networks at:  http://cs231n.stanford.edu/

Office hours

See the website of Renato Campanini