AI Core Subjects (AI4Media Input only so far)

Machine Learning (3IA-UCA)

Syllabus: To be completed.

Offered Courses:

  • Pattern recognitionStatistical machine learning.
    Random variables and vectors. Decision functions. Classification algorithms utilizing decision functions. Classification based on distance. Classification based on decision theory. Principal component analysis. Linear discriminant analysis. Estimation of probability distribution parameters. Analysis of similarity and web graphs. Syntactic pattern recognition. Vector quantization techniques. Programming assignments in C/C++ and MATLAB
    (AUTH, Full course, ECTS:5, Senior undergraduate course, Spring semester, Greek/English, Participation: teleconference/tele-exams/project Lecturer: I. Pitas, Link to course:
  • Advanced Learning: functional, mixed and text data 
    Syllabus: visual mining, time series clustering and embedding, statistical models on graph data, embedding of heterogeneous (structured and unstructured data)(3IA-UCA, ECTS:4, Master 2, Fall semester, English, Lecturers: C. Bouveyron, Frederic Precioso, Marco Winckler,
  • Machine Learning for Visual Data Analysis (QMUL)
  • Statistical Learning in High Dimensions
    Syllabus: statistical analysis of high-dimensional data for classification, clustering, regression, and collaborative filtering, study of sparse methods (lasso, elastic net, sparse Bayesian methods), probabilistic PCA (and mixtures thereof), nuclear norm penalisation. Labs in R and Python.
    (3IA-UCA, ECTS:4, Master 2, Fall semester, English, Lecturers: C. Bouveyron, P.-A. Mattei)
  • Multimedia Classification 
    Syllabus: Introduction, Clustering: data similarity, hierarchical clustering, k-means, Classification: k-NN, Support Vector Machines. Programming assignments in C/C++ and MATLAB.
    (UPB)- (on-demand module, 6-8 hours)
  • Big Data (UPB)
    (Master-level course, taught in English)
  • Data Mining (UPB)
    (Master-level course, taught in English)

Neural networks. Deep Learning

Syllabus: To be completed.

Offered Courses:

  • Deep Learning for Computer Vision (QMUL)
  • Adversarial Learning and Explainable AI(UNIFI)
  • Graph Neural Networks (3IA-UCA, Not available this year, ECTS:4, Master 2, Fall semester, English, Lecturers: M. Gori)
  • Computer Vision and Deep Learning in Practice (UNIFI, Short Course)
  • Deep Learning
    Syllabus: The course aims to provide students with an overview of the main models and applications of deep learning. In particular, the first part of the course will introduce the basic concepts related to deep learning and to the training of artificial neural networks (Backpropagation, Dropout, BatchNorm, …). In the second part, the main types of neural models will be presented. Convolutional Neural Networks, Recurrent Neural Networks, Generative Adversarial Networks, Deep Reinforcement Learning will be introduced. In the final part of the course some applications of deep learning will be presented in the field of computer vision, robotics and natural language processing. Theoretical discussion will be complemented with lab in Python using open-source deep learning libraries. (UNITN,  Full course, ECTS:6, Graduate course, Spring semester, English, Lecturers: Elisa Ricci (
  • Introduction to Deep Learning
    Syllabus: CNN, RNN, attention models for RNN, hyper-parameter tuning, convolutional and variational auto-encoders, Applications to (1) NLP: Word embedding, Sentiment analysis, Naming Entity Recognition, Part of speech tagging, (2) image processing: feature and image extraction, image denoising.
  • Advanced Deep Learning
    Syllabus: Deep neural architectures for multi-modal media data, Deep generative models, Generative Adversarial Networks, Bias mitigation, Deep networks interpretation and explanation, latest research challenges in DL (double gradient descent, exponential learning rate, etc.)
    (3IA-UCA, ECTS:4, Master 2, Fall semester, English, Lecturers: F. Precioso, P.-A. Mattei)
  • Multi-Agent Systems (UPB)
    (Master-level course, taught in English)
  • Evolutionary Computing (UPB)
    (Master-level course, taught in English)
  • Computation for Complex Systems (UPB)
    (Master-level course, taught in English)
  • Optimization for Big Data (UPB)
    (Master-level course, taught in English)
  • Autonomous Agents (UPB)
    (Master-level course, taught in English)

Computer Vision (AUTH)

Syllabus: 2D computer vision: Image acquisition. Edge detection. Image segmentation. Image texture. Image features. Image registration. Image search and retrieval. Image topology. 2D shape description and recognition. 3D computer vision: Mathematical modeling of image formation. Camera calibration. Stereo vision. Depth estimation. Object localization. SLAM. 3D image analysis. Surface geometry. 3D object topology. 3D object landmarks and features. 3D object recognition. 3D object registration. 3D object description. Applications in medical imaging, image retrieval, robotic vision.Video analysis: Moving images. Motion estimation. Object tracking. Video description. Video search and retrieval.

Offered Courses:

  • Advanced Computer Vision
    Syllabus: Image acquisition. Mathematical modeling of image formation. Introduction to image processing and analysis. Camera calibration. Stereo vision. Depth estimation. Object localization. 3D image analysis. Surface geometry. Object topology. Object landmarks and features. Object recognition. Object registration. Object description. Applications in medical imaging, image retrieval, robotic vision.
    (AUTH, Full course, ECTS:7.5, Graduate course, Fall semester, Greek/English, Participation: teleconference/tele-exams/project Lecturer:  I. Pitas,
  • Advanced Computer Vision
    Syllabus: This course will prepare the students in both the theoretical foundations of computer vision as well as the practical approaches to building real Computer Vision systems. This course investigates current research topics in computer vision with an emphasis on recognition tasks and deep learning. We will examine data sources, features, and learning algorithms useful for understanding and manipulating visual data. The goal of this course is to give students the background and skills necessary to perform research in computer vision and its application domains such as robotics. Students should understand the strengths and weaknesses of current approaches to research problems and identify interesting open questions and future research directions. Topics covered will include object detection and segmentation with deep networks, deep generative models for image generation, image captioning, activity recognition, video generation, deep transfer learning and few-shot learning.
    (UNITN,  Full course, ECTS:6, Graduate course, Fall semester, English, Lecturers: Nicu Sebe & Elisa Ricci (,
  • Computer vision and Machine Learning for Autonomous Systems (Short course, ECTS 2, Graduate course, Spring semester, English, Participation: local/exams/project Lecturer:  I. Pitas, AUTH)

Network/web science. Social media analytics (3IA)

Syllabus: Principles of a Web of Linked Data, the RDF Data Model to publish and link data on the Web, the SPARQL query language, Integration of other data formats and sources

Offered Courses:

  • Web of Data
    Syllabus: Principles of a Web of Linked Data, the RDF Data Model to publish and link data on the Web, the SPARQL query language, Integration of other data formats and sources.
    (3IA-UCA, ECTS:4, Master 2, Fall semester, English, Lecturer: C. Faron)
  • Statistical Analysis of Graphs
    Syllabus: Network inference: estimation of network characteristics (diameter, edge conductances, etc.), testing hypotheses regarding graph structure, rumour source detection, network tomography; Network algorithms: distributed learning and optimization, clustering.
    (3IA-UCA, ECTS:4, Master 2, Fall semester, English, Lecturer: K. Avrachenkov)
  • Human-Centered Social Media (IDIAP)
    Syllabus: The course presents a human-centered, multidisciplinary view of social  media. It integrates concepts from media studies, multimedia information systems, and machine learning to understand user motivations and behavior, and analyze content of socio-technical systems like Twitter, Facebook, and YouTube. Students will become familiar with approaches for classification, discovery, and interpretation of social media phenomena.
    (IDIAP, short course for master or early doctoral levels. It can be offered as part of a summer or winter school if AI4Media coordinates that. It does not give any official ECTS credits, taught in English. Lecturer: D. Gatica-Perez,

Human-centered computing (UNITN)

Syllabus: This course explores computing research that relates to, arises from, or deliberately influences emotion. The aim is to identify the important research issues, and to ascertain potentially fruitful future research directions in relation to the multimodal emotion analysis and to human-computer interaction. The course will introduce key concepts, discuss technical approaches, and open issues in the following areas: interaction of emotion with cognition and perception; the role of emotion in human-computer interaction; the communication of human emotion via face, voice, physiology, and behavior; construction of computers that have skills of emotional intelligence; the development of computers that “have” emotion; and other areas of current research interest in the research community.

Offered Courses:

  • Affective Computing (UNITN, Full course, ECTS:6, Graduate course, Fall semester, English, Lecturer:  N. Sebe,

AI and Society Courses (KUL)

  • AI Ethics and Regulation (KUL)
    Syllabus: online materials covering following topics: *Requirements of trustworthy AI: Ethical and legal perspectives, critical discussion and implementation (autonomy and personhood; safety and security; justice; enforcement and regulatory oversight); *Dual Use: ethical backdrop, policies and implementation; *Comparison EU with non-EU perspectives on ethical / trustworthy AI; *Case studies (including media & fake news).
    (KU Leuven, ECTS: 4, Master 2, Spring semester, English, Lecturer: P. Valcke & A. Vedder)
  • IP/IT Law and Emerging Technologies (KUL)
    Syllabus: online materials covering following topics: *emerging legal issues in the fields of intellectual property rights, liability, privacy and data protection in relation to new technologies, in particular Artificial Intelligence, Nanotechnology, Neurotechnology, Biotechnology and Robotics.
    (KU Leuven, ECTS: 3, Master 2, Spring semester, English, Lecturer: D. Burk)

AI4Media Elective Topics

Image Processing and Analysis

Syllabus: Digital image recording/digitization. Image enhancement and filtering. Image restoration. Digital image compression and related standards (e.g. JPEG/JPEG2000). Edge detection algorithms. Image segmentation. Shape description. Programming assignments in C/C++ and MATLAB.

Offered Courses:

  • Pre-processing of Visual Information (UPB)
    (on-demand module, 6-8 hours) Introduction & applications, Image/video representation, Color representation, Point operations, Linear/non-linear filtering. Programming assignments in C/C++ and MATLAB.
  • Multimedia Content Representation (UPB)
    (on-demand module, 6-8 hours) Introduction, Data representation, Content description (color, shape, texture, motion, temporal structure, interest points, audio, text), Normalization, Decorrelation. Programming assignments in C/C++ and MATLAB.
  • Machine Listening for Music and Sound Recognition (IDMT) 
    Syllabus: 1) Introduction to Audio Representations, 2) Introduction to Machine Learning, 3) Music Information Retrieval, 4) Machine Listening for Environmental Sounds (IDMT, 4 lectures + 3 seminars, ECTS: no, graduate student course, Winter semester, English, Participation: teleconference, Lecturer: J. Abe├čer,,

Multimedia Information Retrieval

Syllabus: To be completed.

Offered Courses:

  • Multimedia information retrieval
    Syllabus: The course can include a large part on evaluation and benchmarking of multimedia retrieval (ImageCLEF, MediaEval)

Games (UM)

Syllabus: Ever since the birth of the idea of artificial intelligence, games have been helping AI research progress. Games not only pose interesting and complex problems for AI to solve, they also offer a canvas for creativity and expression which is experienced by their users. Thus, arguably, games are a rare domain where science (problem solving) meets art and interaction: these ingredients have made games a unique and favourite domain for the study of AI. But it is not only AI that is advanced through games; games have also been advanced through AI research. AI has been helping games to get better on several fronts: in the way we play them, in the way we understand their inner functionalities, in the way we design them, and in the way we understand play, interaction and creativity.

Yannakakis and Togelius, Artificial Intelligence and Games, 2018. Springer Nature

Offered Courses:

  • Computational Game Creativity (UM)
    Description: Computational game creativity covers the different creative facets of games (visuals, audio, narrative, game design, level design and gameplay) and attempts to automate one or more of these facets. Autonomous or semi-autonomous computational creators can alleviate the authorial burden of games during development, can lead to interesting and unexpected gameplay experiences and can provide insights on the nature of human creativity. Computational game creativity is positioned at the intersection of developing fields within games research, such as procedural content generation and AI-assisted design, and long-studied fields, such as visual art and narrative.

    The study-unit includes a study of the different game facets from the perspective of human authors, presents the state-of-the art in procedural content generation within games, and connects them to instances of computational or mixed-initiative creativity outside of games (e.g. parametric design in architecture, generative art, procedural audio).

    Syllabus: Introduction, Level & World Generation, Rules and Mechanics, Experience-driven Procedural Content Generation (PCG), Visuals & Audio, Narrative, Mixed-Initiative PCG, Evaluating PCG
  • Player Modeling: From Game Analytics to Affective Computing (UM)
    Description: The primary goal of the unit is to revisit the field of game artificial intelligence (AI) and introduce non-traditional uses of AI in games. A short introduction will be given on AI areas that are currently reshaping the game AI research and development roadmap including procedural content generation, player experience modeling, and AI-based game design. The primary focus of the unit, however, will be on player modeling (spanning from game analytics and game data mining to affective computing methods). Within game data mining, emphasis will be given on state-of-the-art data analytics/mining algorithms and methods for improving the gameplay experience and game development procedures. Within affective computing, emphasis will be given in the phases of emotion elicitation, emotion recognition (feature extraction, feature selection, annotation, classification, regression, preference learning), emotion expression (e.g., facial expression, agent behavioural responses, etc.) and affect-driven adaptation (interaction elements adapt to the user needs/affect).

    Syllabus: Revisiting game artificial intelligence. The role of Player Modeling, Basic data analysis, data preprocessing and descriptive statistics, Classification and prediction, Clustering, Data Visualization, Industrial game analytics – problems and needs, Theories of emotion (affect and cognition), The Affective Loop: key components, Eliciting Emotion (protocols and approaches), Recognizing and Modelling Emotion, The model’s input (Speech, eye gaze, physiology, images, movement/posture), Feature Extraction / Selection, The model’s output (affect annotation / ranks, ratings, ground truth), A taxonomy of modelling approaches, Pattern recognition, Classification, Regression, Preference Learning, Expressing Emotion (via agents and virtual environments), Closing the affective loop: Adaptation via agents and virtual environments, Player Experience Modeling, Popular application domains: computer games, HCI, health etc.
  • Game Artificial Intelligence (UM)
    Description: Game technology incorporates a number of core technical fields that are relevant for modern game development, design and production. Most of these areas are driven by key artificial intelligence (AI) techniques such as expert domain-knowledge systems, search and optimization, and computational intelligence in games. The primary goal of this unit is the understanding, design, implementation and use of basic and nouvelle AI techniques for generating efficient intelligent behaviors in games. The study-unit aims to introduce students to the theory of basic and advanced game artificial intelligence topics and provide hands-on experience on the implementation of popular algorithms on commercial-standard games.

    Syllabus: Introduction to Game AI, Representation and Utility, Play games!, Tree Search and Monte Carlo Tree Search, Play via Supervision, Play via Reinforcements, Play via Evolutionary and Genetic Algorithms, Play for Winning, Diversity and Testing

    Course page:

Autonomous Systems for media production

Syllabus: To be completed.

Offered Courses:

  • Autonomous Systems Perception (AUTH, Full course, ECTS:5, Graduate course, Spring semester, Greek/English, Participation: teleconference/tele-exams/project Lecturer: I. Pitas,


Offered Courses:

  • Chaos and Fractals (UPB)
    (Master-level course, taught in English)
  • Computation for Complex Systems (UPB)
    (Master-level course, taught in English)