Artificial Intelligence and Multimedia Technologies

The course explores artificial intelligence technologies and machine learning for multimedia processing and analysis and compression technologies, search and semantic description of multimedia information. More specifically, the course covers: 

Analysis of the concepts of sampling and digital representation of multimedia information. Description of traditional methods of processing and analyzing multidimensional signals (Fourier transformation, filters (linear, non-linear), geometric transformations). Deep dive intomachine learning methods and especially in supervised, unsupervised and semi-supervised learning methodologies for multimedia analysis. Description of algorithms and unsupervised learning technologies with emphasis on geometry and minimum distance methods. Reference to the structure of the simple neuron and the corresponding optimization algorithms for finding and estimating its parameters. Description of linear regression for the prediction and categorization of multimedia information. Introduction to the structures of multilayer neural networks and the corresponding optimization methods for estimating their parameters. Description of the convolution operation and analysis of new machine learning methods that are based on the convolution operation, convolutional neural networks. Introduction to deep machine learning structures and learning based on self-encoding. Description of competitive learning technologies and applications in the synthesis of multimedia data. 

The course is taught using Python programming environment for management and libraries for processing multimedia data and applying machine learning. Technologies for image and video compression, such as JPEG, MPEG, and standards for semantic description of multimedia (MPEG-7) are presented.. 

You can find the course page here. 

Professors: 

N. Doulamis, Professor NTUA 

A. Doulamis, Associate Professor NTUA