Machine Learning and Pervasive Computing (Winter 2014/2015)

Details

Workload/ECTS Credits:	180h, 5 ECTS
Module:	M.Inf.1223: Spezielle fortgeschrittene Aspekte der Computernetzwerke
Lecturer:	Stephan Sigg
Teaching assistant:	--
Time:	Wednesdays, 10.15 - 11.45.; Exercise: 16:15 - 17:45 (bi-weekly)
Place:	IfI 3.101
UniVZ	[1]

Course Overview

The course will address selected topics in Pervasive Computing. This semester the main focus of the lecture will be on Machine learning and activity recognition from sensor-data. In addition, other fields of Pervasive Computing are covered to provide students with a good overview on current advances and research challenges. Depending on the interest of the students, the emphasis on these additional topics may differ.

• Course topics
  • Introduction to Machine learning
  • Supervised and Unsupervised learning
  • Features and feature extraction
  • Feature subset selection
  • Performance metrics
  • Polynomial curve fitting
  • Support Vector Machines
  • Artificial Neural Network learning
  • Clustering (k-means)
  • Dimensionality reduction
  • Anomaly detection
  • Recommender systems

Schedule

• Wednesday, October 22nd, 2014, 10.15 - 11.45: Lecture 1
  • Organisation (pdf)
  • Introduction (pdf)
• Wednesday, October 29th, 2014, 10.15 - 11.45: Lecture 2
  • Classification and Evaluation (pdf)
• Wednesday, November 5th, 2014, 10.15 - 11.45: Lecture 3
  • Simple supervised learning (pdf)
• Wednesday, November 5th, 2014, 16.15 - 17.45: Exercise 1
  • Assignment 1 (pdf)
• Wednesday, November 12th, 2014, 10.15 - 11.45: Lecture 4
  • Random search (pdf)
• Wednesday, November 19th, 2014: No lecture
• Wednesday, November 26th, 2014, 10.15 - 11.45: Lecture 5
  • Bayesian networks and Naive Bayes (pdf)
• Wednesday, November 26th, 2014, 16.15 - 17.45: Exercise 2
  • Assignment 2 (pdf)
• Wednesday, December 3rd, 2014: No lecture
• Wednesday, December 10th, 2014, 10.15 - 11.45: Lecture 6
  • Decision Tree (pdf)
• Wednesday, December 17th, 2014, 10.15 - 11.45: Lecture 7
  • K-nearest neighbour (pdf)
• Wednesday, December 17th, 2014, 16.15 - 17.45: Exercise 3
• Wednesday, January 7th, 2015, 10.15 - 11.45: Lecture 8
  • Support Vector Machines (pdf)
• Wednesday, January 14th, 2015, 10.15 - 11.45: Lecture 9
  • Artificial neural networks and self organizing maps (pdf)
• Wednesday, January 14th, 2015, 16.15 - 17.45: Exercise 4
• Wednesday, January 21st, 2015, 10.15 - 11.45: Lecture 10
  • Hidden markov models and conditional random fields (pdf)
• Wednesday, January 28th, 2015, 10.15 - 11.45: Lecture 11
  • Dimensionality reduction and unsupervised learning (pdf)
• Wednesday, January 28th, 2015, 16.15 - 17.45: Exercise 5
• Wednesday, February 4th, 2015, 10.15 - 11.45: Lecture 12
  • Anomaly detection, online learning and recommender systems (pdf)

Requirements

• Each participant is required to attend and actively participate in the exercises.
• Oral examination at the end of the semester

Reading List

• Pervasive and Ubiquitous Computing
  • [Marc Weiser on challenges in Ubicomp]
  • [Recent challenges with new devices and interaction]

• Activity recognition
  • [Activity recognition in Pervasive Computing]