Opening (Monday 14:00-14:30)
Opening of the school. Logistic information.
Student sessions (Monday 14:30-17:30 and Tuesday 9:30-11:00) PhD students
After a successful PhD student session which took place in Napoli, the Krakow school will also implement this idea. We are going to start from sessions where the school participants will present theirs work. Each participant is encouraged to present a 10 minute talk about his/her current work. This presentation will be followed by 5 minutes for questions and clarifications.
Machine Learning (Tuesday 11:30-17:30 ) Pedro Casas
Two lectures of 1.5 hours: An Introduction to Machine Learning in Networking.
* What is Machine Learning (ML) and why applying ML in Networking?
* ML as a discipline vs ML as a tool to solve complex problems.
* General overview on ML: supervised, unsupervised, semi-supervised, ensemble learning.
* The paramount role of features' extraction and selection.
* Some brief examples of ML in Networking (QoE, Traffic Classification, Network Security).
One slot of 1.5 hours for Lab exercises: Machine Learning in Networking using the WEKA Toolbox.
* Applying different ML techniques to Traffic Classification using the WEKA toolbox (very popular Java-based toolbox for ML and Data Mining from the University of Waikato, New Zealand).
The ML part of the lectures will be mainly based on the text book "Pattern Classification" from R.O. Duda; the examples of ML in Networking will be based on some selected papers; the lab will be based on the WEKA software (http://www.cs.waikato.ac.nz/ml/weka/).
Traffic Classification (Wednesday 9:30-15:30) Marco Mellia
Two lectures of 1.5 hours:
* Traffic Classification - Mellia-Classification-KrakowTMA(.pptx) - from port based to Deep Packet Inspection to behavioral classifiers - Mellia-Classification-KrakowTMA
* Examples of advanced behavioral classifiers: the case of Skype and UDP traffic
* Data analysis: Tools to dig into data and how to manage logs
* YouTube example: experiences from studying youtube architecture and usage
One slot of 1.5 hours of Lab exercises:
* Traffic trace analysis: students will be provided some example of traces and tools to run some traffic characterization and classification example.
* Homework: This is the description of the homework students have to prepare.
Two lectures of 1.5 hours:
The first lecture "Quality of Experience and Web QoE" by Raimund Schatz
* Quality of Experience in general and how to measure it
* Generic Relationships between QoE and QoS
* Web QoE
The second lecture"YouTube QoE - Crowdsourcing, Modeling, Monitoring" by Tobias Hossfeld
* Crowdsourcing experiments for YouTube QoE: Introduction to Crowdsourcing; Design and Implementation of Crowdsourcing tests; Reliability Analysis of Subjective Results
* Modeling of YouTube QoE: Derivation of Key Influence Factors; QoS-QoE Mapping Functions; Comparison of Crowdsourcing and Lab Results
* Monitoring YouTube QoE: Application-Level Measurements of YouTube; Network-Level Measurements of YouTube; Evaluation of In-Network Monitoring Approaches
One slot of 1.5 hours of Lab exercises "Design and Analysis of Crowdsourcing Experiments" guided by Tobias Hossfeld and Raimund Schatz
* Lab exercise consists of three different parts, here is the needed data file.
* Exercise A. Design of Reliable Crowdsourcing Tests
* Exercise B. Statistical Reliability Analysis of User Tests. Excel data (if not using Matlab)
* Exercise C. Curve Fitting for QoS-QoE mapping functions
Three slots of 1.5 hours of Lab exercises:
* QoE definition and its consequences
* Different experiments types
* How to prepare subjective experiment
* Problems of subjective experiment preparation for video quality evaluation.
- Choosing sequences.
- Compression - basic features.
- Network streaming emulation - Sirannon.
- Packet losses.
- PVSes generation.
- Sequences revision.
- Subjective experiment running.
* Data analysis.
* JEG (Joint Effort Group).
Exercise - plan a lab subjective experiment for an application different from video or audio streaming. The document should address at least all problems mentioned at slide 44. Each part of the experiment preparation should be described shortly.
Deadline 15 March Deadline 23 March (hard deadline).
Test and Closing (Friday 11:30-13:00)
Test and closing