Courses – Fakultet tehničkih nauka u Novom Sadu

Subject: Machine learning 1 (17.EK466)

Native organizations units: No data

Type of studies	Title

General information:

Category	Theoretical-methodological
Scientific or art field	Telecommunications and Signal Processing
Interdisciplinary	No
ECTS	6

Educational goal:

Introduction to basic machine learning concepts and algorithms including theoretical foundations, analysis and practical applications. The course provides understanding of the relevant supervised and unsupervised learning approaches and offers best practices and operational advises on their implementation.

Educational outcome:

Students will be able to identify and exemplify machine learning problems. They will be able to interpret and analyze machine learning algorithms, implement them (in Python), and evaluate algorithms' performance. Students will know how to combine the algorithms and compose workflows from data preprocessing to performance validation step. Gaining experience on how to overcome common implementation problems (accuracy, computational cost, overfitting, regularization).

Course content:

Introduction and basic concepts. Machine learning system's components. Learning types. Approaches to ML. Different machine learning problems. Fundamental concepts: cost-functions, overfitting, regularization, cross-validation, bias-variance trade-off, curse of dimensionality. Supervised learning (Bayesian decision theory; quadratic classifiers; density estimation: parametric (Maximum Likelihood and Bayesian estimation) and non-parametric (kernel density estimation, kNN); linear and logistic regression; linear discriminative functions; neural networks; support vector machines) Unsupervised learning (k-Means Clustering; Hierarchical clustering) Dimensionality reduction: PCA , LDA.

Teaching methods:

Lectures, computer lab sessions (Python and other appropriate programming envionments), homework, consultations, active learning, project and research based learning, workshops.

Literature:

Authors	Title	Year	Publisher	Language
Kevin Murphy	Machine Learning: A Probabilistic Perspective	2012	MIT Press	English
Bishop, C.M.	Pattern Recognition and Machine Learning	2006	Springer, New York	English
Crnojević, V.	Prepoznavanje oblika za inženjere	2014	Fakultet tehničkih nauka, Novi Sad	Serbian language
Richard O. Duda, Peter E. Hart, David G. Stork	Pattern Classification, 2nd Edition	2001	Wiley	English

Knowledge evaluation:

Course activity	Pre-examination	Obligations	Number of points
Homework	Yes	Yes	5.00
Project	Yes	Yes	30.00
Homework	Yes	Yes	5.00
Written part of the exam - tasks and theory	No	Yes	50.00
Lecture attendance	Yes	Yes	3.00
Computer exercise attendance	Yes	Yes	2.00
Homework	Yes	Yes	5.00