
Course unit
HUMAN DATA ANALYTICS
INP7080694, A.A. 2017/18
Information concerning the students who enrolled in A.Y. 2017/18
ECTS: details
Type 
ScientificDisciplinary Sector 
Credits allocated 
Core courses 
INGINF/03 
Telecommunications 
6.0 
Course unit organization
Period 
Second semester 
Year 
1st Year 
Teaching method 
frontal 
Type of hours 
Credits 
Teaching hours 
Hours of Individual study 
Shifts 
Lecture 
6.0 
48 
102.0 
No turn 
Start of activities 
26/02/2018 
End of activities 
01/06/2018 
Examination board
Board 
From 
To 
Members of the board 
1 a.a 2017/2018 
01/10/2017 
30/09/2018 
ROSSI
MICHELE
(Presidente)
ERSEGHE
TOMASO
(Membro Effettivo)
BADIA
LEONARDO
(Supplente)
LAURENTI
NICOLA
(Supplente)
TOMASIN
STEFANO
(Supplente)
ZANELLA
ANDREA
(Supplente)

Prerequisites:

A basic course on probability theory is recommended. A basic programming course is also advised for a correct comprehension of the material. 
Target skills and knowledge:

The skills that will be acquired are related to the analysis of biosignals and in general to those signals that are generated by human activity at large. In particular, in the course the students will learn to:
 Deal with quasiperiodic signals, extracting some recurrent structure from them (denoising, segmentation): some key signals that will be studied are the electrocardiogram signal (ECG) and motion data from wearables;
 Extract relevant features from biometric datasets;
 Apply classification (clustering) algorithms on biosignals to classify users or build dictionaries for the compact and accurate representation of the data;
 Use unsupervised learning methods, base on artificial neural networks, to perform vector quantization (dictionary learning);
 Use statistical structures to track correlation in the data such as Bayesian networks and Hidden Markov Models;
 Use supervised neural networks as mapping tools to learn patterns and solve classification problems. 
Examination methods:

The exam will be a written project where the student will have to solve a selected classification problem for a given dataset. The student will have to work on the code (software), produce and discuss the results. 
Assessment criteria:

The following evaluation criteria will be taken into account:
 Oral presentation skills;
 Quality of written project: clarity of exposition, mathematical rigor;
 Quality of the obtained results;
 Originality of the adopted techniques;
 Problem complexity. 
Course unit contents:

Part I – Introduction
 Intro: course outline, grading rules, office hours, etc.
 Applications: health, activityaware services, security and emergency management, authentication systems, analyzing human dynamics
Part II  Tools
 Vector quantization (VQ):
 Aims, quality metrics
 Unsupervised VQ algorithms:
 SelfOrganizing Maps (SOM), Time AdaptiveSOM (TASOM)
 Gas Neural Networks (GNG)
 Deep Neural Networks (DNN)
 Neural networks brief: concept, examples, training
 Convolutional Neural Networks (CNN): structure, training
 Sequential Data Analysis:
 Hidden Markov Models (HMM):
 Maximum Likelihood for the HMM
 Forwardbackward algorithm
 Sumproduct algorithm
 Viterbi algorithm
 Wald's sequential decision theory (iid case)
Part III – Applications (using the tools of Part II)
 Human Activity Learning
 Activities & sensors: definitions, classes of activities
 Features: sequence features, statistical features, spectral features, activity context features
 Activity recognition: activity segmentation, sliding windows, unsupervised segmentation, performance measures and results
 Biometric Data Processing:
 Quasiperiodic physiological signals
 Dictionary learning & compression algorithms
 Inertial signals: identity recognition
 Inertial sensors and cameras: gait tracking & profiling
 Voice Recognition:
 Deep neural networkHMM hybrid system
 Architecture, training, performance evaluation 
Planned learning activities and teaching methods:

The course will be based on frontal lessons that will be given by the Professor in charge. The final project will also serve to put the techniques discussed in the classes into good use, and verify their effectiveness when applied to real signals. 
Additional notes about suggested reading:

The course material will consist of:
1) book chapters (different and selected books will be used, depending on the specific topic);
2) scientific papers;
3) slides.
All the written material will be in English. The material of the above points 1), 2) and 3) will be made available (password protected) through the course site. 
Textbooks (and optional supplementary readings) 

Bishop, Christopher M., Pattern recognition and machine learningChristopher M. Bishop. New York: Springer, .

Cook, Diane J.; Krishnan, Narayanan C., Activity learningdiscovering, recognizing and predicting human behavior from sensor dataDiane J. Cook, Narayanan C. Krishnan. Hoboken: NJ, Wiley, 2015.

Jeremy Watt, Reza Borhani, Aggelos K. Katsaggelos, Machine Learning Refined: Foundations, Algorithms, and Applications. : Cambridge University Press, 2016.

Goodfellow, Ian; Courville, Aaron, Deep learningIan Goodfellow, Yoshua Bengio and Aaron Courville. Cambridge: MA [etc.], MIT Press, 2016.


