MACHINE LEARNING FOR PATTERN RECOGNITION (2st MODULE)
Learning outcomes of the course unit
The objective of this module is to provide students with the theoretical basis and practical knowledge of some relevant machine-learning algorithms, aimed at classifying data.
The methods described in the course will allow students to:
- learn and use inductive-learning algorithms
- learn and use neural nets and other algorithm classes for the supervised classification of data
- learn and use the main supervised and unsupervised clustering algorithms
Entry-level courses in linear algebra and probability theory, such as those normally offered in the corresponding 3-year Laurea course, are necessary pre-requisites for this course.
Course contents summary
Part 1: Introduction
Lesson 1: How to set up a machine learning experiment
Lesson 2: Learning-based classification
Part 2: Neural networks
Lesson 3: Introduction to neural networks
Lesson 4: Supervised and unsupervised learning
Lesson 5: Supervised learning: the Backpropagation algorithm
Lesson 6: Unsupervised learning and clustering
Lesson 7: Kohonen's self-organizing maps (SOM)
Lesson 8: Learning Vector Quantization
Part 3: Other learning-based classifiers
Lesson 9: Support Vector Machines
Lab 1: WEKA
Lab 2: Classifiers in WEKA: Multi-Layer Perceptrons
Lab 3: SOM-based clustering
 C. W. Therrien, "Decision, estimation and classification" Wiley, 1989
 C. M. Bishop "Pattern Recognition and Machine Learning", Springer, 2006.
 R O Duda, P, E. Hart, D. G. Stork, "Pattern classification", 2nd Ed., Wiley, 2001
Classroom teaching, 18 hours.
Labs, 6 hours.
Homework regularly assigned.
Assessment methods and criteria
A practical project will be assigned, whose results will be presented and discussed by the student both as a written report and as an oral presentation.
By appointment (Scientific Complex, Building 1, floor 2, email stefano.cagnoni[AT]unipr.it).