Programma en vakken Cognitive Science and Artificial Intelligence
Onderzoek verschillende aspecten van intelligentie, zoals redeneren, leren, perceptie, communicatie en samenwerking vanuit zowel menselijk als technisch perspectief.
Deze twee-jarige opleiding bestaat uit 120 ECTS, waarvan:
- 5 Artificial Intelligence vakken (30 ECTS)
- 5 Cognitive Science vakken (30 ECTS)
- 3 Research Skills vakken (9 ECTS)
- 1 keuzevak (6 ECTS)
- Onderzoeksstage (15 ECTS)
In het eerste semester van het tweede jaar loop je (de verplichte) stage.
- Project en masterthesis (30 ECTS)
Je sluit het programma af met een geavanceerd interdisciplinair onderzoeksproject dat resulteert in een scriptie.
- Extra-curriculair: studentassistent
Het programma heeft één startmoment: eind augustus.
Inhoud van het programma
Vakken per jaar | semester
AI = Artifical Intelligence course | CS= Cognitive Science course | RS = Research Skills course
Kies één van de volgende vakken:
Artificial Intelligence vakken
Core Topics in Artificial Intelligence
The general aim of this course is for students to become familiar with advanced topics in artificial intelligence (AI), specifically relating to social signal processing and social network analysis for modeling social behavior and human-machine interaction using human-centric AI methods.
Complex systems pervade our everyday lives with everything from cells to civilizations (and beyond) exhibiting dynamic and complex behavior. These systems have many components that interact with each other in ways that give rise to emergent and self-organized patterns of behavior that change over time. For example, complex systems approaches have been used to detect effective sport and business team coordination to improve performance, predict dramatic shifts in the stock market/economy, understand the dynamics of human cognition and behavior, model the process of human change during psychotherapy, and differentiate healthy vs unhealthy physiological patterns.
This course will focus on the theory and mathematical methods for understanding complex systems as they apply to cognitive and data scientists. It will cover theories of change, coordination, emergence, self-organization, phase transitions and critical instabilities. The methods utilized will fall under the general categories of dynamical systems modeling and time-series analysis to include difference/differential equation modeling, fractals, multi-fractals, phase space reconstruction, recurrence quantification, surrogate analysis, wavelet-transform and cross-coherence, and information theoretic measures (e.g., entropy, mutual information). Additionally, the practical emphasis of the course will be on applying the theory and methods to a data set.
This course requires R and/or Python programming skills.
Deep Reinforcement Learning
Reinforcement Learning allows autonomous agents with complex goals to learn from trial-and-error interactions with their environment using delayed returns (rewards and punishments).
The course introduces the basics and theory of reinforcement learning, starting from Markov Decision Processes (MDPs) and traditional algorithms like TD-and Q-learning. The theory underlying the field will be presented in relation to current knowledge in neuroscience on rewards-based learning of behavior in biological agents.
The lectures will then focus on the recent improvements to the algorithms (Mnih et al., 2013, and subsequent work) that made reinforcement learning possible using deep neural networks, addressing the limitations of previous algorithms.
The course will present an overview of the major state-of-the-art deep reinforcement learning algorithms and the current research topics in the field will be introduced and discussed. The course has an emphasis on applications, so while the students will be required to be able to implement the most commonly used algorithms in a classroom setting, they will also be taught how to adapt existing optimized implementations for use in practice.
The goal of this course is to teach the students theories and techniques for creating natural behavior in virtual environments, and how to implement artificial behavior in virtual environments.
Artificial intelligence in games allows a computer to play a game, or be an integral part of a world in a game. The course discusses how such artificial intelligence is created, what techniques are used in state-of-the-art games, and which techniques will be used in the future of games.
Advanced Deep Learning
In the course Advanced Deep Learning students are trained to perform research in advanced deep learning. Two recent topics in advanced deep learning are offered from which each student selects one.
The course is research-oriented in the sense that lectures have the form of sessions in which individual students (or student groups, depending on the total number of students enrolled in the course), give a presentation on their progress. The instructors and fellow students provide feedback and provide help in case of obstacles.
Cognitive Science vakken
Core Topics in Cognitive Science
The general goal of this course is to provide students with knowledge of current advances in the field of Cognitive Science thus ensuring a common ground for all students in the program.
Areas and topics covered by this course address the fundamental properties of brain computation, the relationship between neural structure and cognitive function, the organization of the computational processes in the overall hierarchy and how the brain uses computation to manage behavior.
The cognitive functions discussed include learning, perception, attention, memory, pattern recognition, categorization, and executive functions.
Cognitive Models of Language Learning
This course is about using formal models for studying human language learning.
In the first part of the course, the focus will be on the high-level objectives of studying human language, the general properties of computational models of language learning, and the most common frameworks for developing them.
In the second half of the course, the focus will be on the cognitive processes involved in human language acquisition including speech segmentation, the association of words to meanings, learning language structure, and formation of linguistic categories, and analyze a number of computational models of each of these aspects.
Risk Communication and Uncertainty
Most decisions are made with uncertain information. This course introduces students to the foundations of probability, the human perception and understanding of risk and uncertainty, and the contemporary debate surrounding the ability of humans to make rational decisions under uncertainty.
In short, this course will equip students with an understanding of risk and uncertainty, how to convey these risks and uncertainties in a transparent way, and how these issues relate to the contemporary debate on human rationality.
Brain Computer Interfacing
Brain-Computer Interfaces (BCIs) are systems that translate human brain activity into a command for an interaction application. This course gives an introduction to the BCI technology and approaches it from a theoretical and practical perspective.
On the theoretical side, students will gain knowledge about the key definitions, history, algorithms and techniques, applications and shortcoming of the current BCI systems.
On the practical side, the course focuses on basic bio-signal analysis, BCI modeling, and application in scientific research. Students will gain hands-on experience with EEG systems (such as g.Nautilus, BITalino, Emotiv, B-Alert, etc.) and learn to use common signal processing tools/softwares such as EEGLab, BioSig and MATLAB to analyze and model the EEG data.
Bayesian Models of Cognitive Processes
How can we understand intelligent behavior as computation? This course will introduce probabilistic cognitive modeling through Bayesian probabilistic programs and will explore the probabilistic approach to modeling human and artificial cognition. Examples will be drawn from areas including concept learning, causal reasoning, social cognition, and language understanding.
Research Skills vakken
In this course, we discuss technologies and methods that make it possible to glean valuable insights into large collections of data. We consider a variety of problem settings where big data is encountered and discuss which techniques are the most appropriate for each setting. We begin by presenting some well-known techniques to handle the data in a distributed way, thus speeding up the process significantly. We then discuss the online data streaming problem, in which data must be processed instantly and results produced quickly, while relying on historical data that, due to its volume, cannot be reanalyzed on the spot. In particular, we discuss methods that are capable of quickly producing reliable approximate results when the exact results cannot be computed in a reasonable amount of time.
Spatiotemporal Data Analysis
Spatial data specify “where” and temporal instances specify “when” data is collected. The demand for spatiotemporal analysis is increasing due to the rapid growth and widespread collection of spatiotemporal data across various disciplines. Spatiotemporal analysis can illuminate any unusual patterns and interesting information or allow the study of persistence of patterns over time. This course is an introduction to the challenges and techniques to analyze spatiotemporal data and aims to provide students with the fundamental knowledge of spatiotemporal modelling.
This course provides an introduction to theoretical and practical aspects of digital image analysis including the fundamentals of image formation, image representation and a broad range of basic image processing techniques and algorithms. The course explores different types of image representations, how to enhance image characteristics, image filtering and how to reduce the effects of noise in an image. It also introduces different methods used to extract features and objects in an image. This course will also explore machine learning for image analysis. Machine learning and deep learning is assumed knowledge. Lectures will be complemented with interactive demonstrations and hands-on exercises to provide participants with practical experience in processing images.
Students who have not followed a course in Deep Learning in their prior studies will be instructed to follow the course on Deep Learning offered in the MSc Data Science and Society program in the third block of the academic year (i.e., the first half of the second semester). This way, they will be able to gain the knowledge that is needed in the second part of the course Deep Reinforcement Learning.
Students who have previously followed a course in Deep Learning can choose an elective course from the list of restricted electives below:
- Deep Learning
Deep Learning revolutionized machined learning by yielding the best performance in a large variety of application domains such as speech recognition, image recognition, object detection, drug discovery and genomics. This course provides students with the understanding and skills to apply deep learning to signals, images, videos and textual sources. The course includes a training to run deep learning algorithms on special hardware.
- Natural Language Processing
NLP comprises a vast collection of tasks, algorithms, and theoretical frameworks that, at different levels of linguistic analysis, aim at making human language understandable to computers. In order to build working computer systems that are able to automatically process natural language, it is essential to have a thorough understanding of how these ingredients work. During the course, students acquire this knowledge through the theoretical study of the techniques and practical experiences with basic language processing systems.
- Data Science: Sustainability, Privacy and Security
Students familiarize themselves with key privacy concepts, as well a data protection law and ethics concepts and principles relevant for data science. They learn to identify the challenges that different data science scenarios pose to the legal and ethical principles.
In this course a number of important concepts and debates in moral philosophy are discussed. Students in the course are taught to reflect critically on the central theories of normative ethics, moral responsibility and moral psychology.
In onze onderwijscatalogus vind je de uitgebreide beschrijving van de vakken en boeken die je nodig hebt.
Let op: programma's kunnen soms nog wat wijzigen. Zoek daarom bij de start van je opleiding het actuele programma op in OSIRIS Student.
Wil je meer weten over deze masteropleiding?
Kom kennismaken op een van onze voorlichtingsevenementen
Cognitive Science and Artificial Intelligence in het kort
- Bestudeer kunstmatige intelligentie in relatie tot menselijke cognitie en niet alleen vanuit een technisch perspectief.
- Combineer kennis van menselijke cognitie en kunstmatige intelligentie met technieken als advanced deep learning, deep reinforcement learning en Bayesian modeling.
- Je docenten zijn experts op het gebied van kunstmatige intelligentie, cognitiewetenschap en data science. Zij leren jou de technische vaardigheden om modellen van intelligent gedrag te ontwerpen en te testen.
- Binnen het domein van AI ligt de focus op computational cognition, hybrid intelligence en human-AI interfacing.
- Sterke nadruk op onderzoek: vanaf het begin van je studie word je aangemoedigd om als onderzoeksassistent bij te dragen aan het onderzoek van de faculteit.
Geïnteresseerd in de MSc Cognitive Science and Artificial Intelligence?
Check je toelaatbaarheid en de deadlines voor aanmelden