Courses of Academic Year 2015/2016

A preliminar list of the didactic offer of the ICT doctorate follows. The list will be completed and updated during the year.
Seminars on Bibliometrics and research Evaluation will be offered by UNIMORE for all the doctorate courses.

Handwritten Text Recognition

Lecturer: Moisés Pastor i Gadea

Schedule:

November Monday 7th: 15:00 - 18:00 (3 hours), room P2.4
November Tuesday 8th: 9:00 - 12:00 (3 hours) and 14:00 - 17:00 (3 hours), room P2.4
November Wednesday 9th: 9:00 - 12:00 (3 hours) and 14:00 - 17:00 (3 hours), room P2.2
November Thursday 10th: 10:00 - 11:00 exam, room P2.4

Program:

Part I – Introduction to the handwriting recognition problem
Part II - Acquisition and preprocessing of text images
Part III - Cursive handwriting text recognition (HTR)
Part IV - Optical Models for HTR

Exam:
Not yet defined

CFD: 5

Hidden Markov Models and Selected Applications

Lecturer: Jon Ander Gómez Adriá
Jon Ander Gómez is the Vice-Dean for Lifelong Learning in the School of Informatics of the UPV, and the academic director of the Master in Big Data Analytics. His main research area is Pattern Recognition and in particular Automatic Speech Recognition

Schedule:

October Monday 24th: 11.00-13.00 + 15.00-17.00, room M2.1
October Tuesday 25th: 10.00-13.00 + 15.00-18.00, room M1.8
October Wednesday 26th: 11.00-13.00, room P0.2 + 14.00-18.00, room P2.1

Program:

Module 1.
   - Definition of Markov models
   - Explanation of Markov models and the particular case of Hidden Markov Models (HMM)
Module 2.
   - Discrete HMM
   - Simple example with small data.
   - Training algorithms: Viterbi and Baum-Welch
Practice A
   - Language detection in texts
   - Python code provided
   - Simple case for testing
   - Exercises
       + Changing parameters to see the effects
       + Introduce more than two languages
Module 3.
   - Semi-continuous HMM
   - Clustering algorithms
   - Training algorithms
Practice B
   - Automatic labeling/segmentation of Speech or Handwritten Text by applying clustering and generating codebooks for representing the real valued vectors.
Module 4.
   - Continuous HMM
   - Gaussian Mixture Models
   - Traning algorithms, version adapted to real valued vectors.
Practice C
   - Automatic labelling/segmentation of Speech or Handwritten Text using GMMs for computing the emission probabilities of HMM states

Exam:
Test

CFD: 5

Data Analytics and Visualization

Lecturer: DBGroup UNIMORE & CINECA

Schedule:
19 - 22 September 2016 @ Ingegneria ""Enzo Ferrari"

Program:
Additional information: http://dbgroup.unimo.it/bigdata_settembre_2016

Exam:
Not yet defined

CFD: 6