Course Code

MDA-220

Semester

1st Semester

ECTS Credits

7,5

Type of Course

Mandatory

Machine Learning: Methods and Algorithms

Objective

The course aims at familiarizing the audience with fundamental machine learning techniques and algorithms that cover the spectrum of diverse machine learning applications (supervised / unsupervised learning). The expected learning outcome of the course includes knowledge of the basic machine learning methods and gaining of experience in implementing and using them effectively. It also includes the critical ability for the choice of an appropriate methodology for each distinct machine learning problem, along with the deep understanding of its advantages and weaknesses.

 

Course Contents

Introduction to Machine Learning

Types of Machine Learning, Training Methods, Accuracy Metrics, Prediction, Classification.

Maximum Likelihood Classifiers

Bayesian Concept Learning, Likelihood, Model fitting, Naive Bayes Classifier, Bayesian Networks.

Decision Trees

Tree representation, Hypothesis Space Search, Information Gain, ID3 Algorithm, C4.5 Algorithm.

Ensemble Learning and Boosting

Adaboost Algorithm, Random Trees, Combinations of Classifiers.

Gradient Descent for Prediction / Classification

Linear Regression, Logistic Regression, Support Vector Machines. Stochastic Version.

Support Vector Machines

Linear / Non-Linear Classification, Kernel Functions, Multiclass Classification.

Instance-Based Learning

k-Nearest Neighbors Algorithm, Locally Weighted Regression, Traning Examples Selection, RBF Networks.

Programming for Machine Learning in Python

Numpy Library, Visualization with the Matplotlib Library.

Application of Machine Learning in Python

scikit-learn Library.

Weka

Graphical and Programming Environment, Case Studies, Experimentation

RapidMiner

Graphical and Programming Environment, Case Studies, Experimentation


Recommended Readings

  • T. Mitchell. Machine Learning. McGraw-Hill (International Edition), 1997.
  • C. Bishop. Pattern Recognition and Machine Learning. Springer, 2007.
  • K. Murphy. Machine Learning: A Probabilistic Perspective. MIT Press, 2012.