6 ECTS credits
160 u studietijd
Aanbieding 1 met studiegidsnummer 4019816ENR voor alle studenten in het 1e semester met een verdiepend master niveau.
The aim of the course is to cover deep learning algorithms, architectures and systems, and to present applications in various modern data processing and analysis tasks.
The course content is divided in the following parts.
Part I presents concepts around deep neural networks for supervised learning (regression and classification), including forward- and back-propagation, various modern gradient descent based optimization algorithms (e.g., RMSProp, Adam), loss functions and regularization methods. This part covers various deep neural network architectures including fully-connected neural networks, convolutional neural networks, recurrent neural networks, attention mechanisms and graph convolutional neural networks. Applications of such models in computer vision, natural language processing and data mining are also discussed.
Part II presents deep learning models for unsupervised learning, including autoencoders and their regularized and denoising versions. This part also covers deep generative models such as variational autoencoders (VAEs) and generative adversarial networks (GANs).
Part III discusses advanced topics in deep learning including explainability, learning from multimodal data, and learning from few examples.
The goal of this course is to introduce the various concepts, methods, and technologies relevant for the design of deep learning methods for modern data generation, processing and analysis. The students will have the opportunity to follow a set of lectures, implement the concepts during lab sessions in Python and practically deploy these concepts in the form of a project.
At the end of this course, the student will have developed a deep knowledge and understanding in state-of-the-art concepts and technologies in deep learning for supervised and unsupervised learning tasks. The student will be able to formulate, grasp, and analyse various deep learning models and architectures, and to address various data generation, processing and analysis tasks. The student will understand the basic coding technologies in deep learning (including Python libraries, PyTorch, etc.) and will be able to practice these tools in the form of practical sessions. The student will be able to investigate how the acquired theoretical and practical knowledge can be applied to address a practical machine learning problem in the form of a project.
This course contributes to the following programme outcomes of the Master in Applied Computer Science:
MA_A: Knowledge oriented competence
1. The Master in Engineering Sciences has in-depth knowledge and understanding of exact sciences with the specificity of their application to engineering
3. The Master in Engineering Sciences has in-depth knowledge and understanding of the advanced methods and theories to schematize and model complex problems or processes
4. The Master in Engineering Sciences can reformulate complex engineering problems in order to solve them (simplifying assumptions, reducing complexity)
6. The Master in Engineering Sciences can correctly report on research or design results in the form of a technical report or in the form of a scientific paper
8. The Master in Engineering Sciences can collaborate in a (multidisciplinary) team
11. The Master in Engineering Sciences can think critically about and evaluate projects, systems and processes, particularly when based on incomplete, contradictory and/or redundant information
MA_B: Attitude
12. The Master in Engineering Sciences has a creative, problem-solving, result-driven and evidence-based attitude, aiming at innovation and applicability in industry and society
MA_C: Specific competence
18. The Master in Applied Computer Science is able to design and use systems for efficient storage, access and distribution of digital information
19. The Master in Applied Computer Science has knowledge of and is able to use advanced processing methods and tools for the analysis of (big) data in different application domains.
De beoordeling bestaat uit volgende opdrachtcategorieën:
Examen Andere bepaalt 100% van het eindcijfer
Binnen de categorie Examen Andere dient men volgende opdrachten af te werken:
The final exam will be a written evaluation, where the students will address theoretical questions, will be asked to define optimization functions, algorithms and model architectures that solve specific deep learning problems. The project will examine the students’ involvement in the seminar sessions, evaluate their in-depth understanding of deep learning algorithms, models and systems, and assess their practical skills in real-life deep learning tasks.
The final grade is composed based on the following examinations: (1) the result of the final exam, which determines 70% of the final mark; and (2) the result of a project work, which determines 30% of the final mark.
Deze aanbieding maakt deel uit van de volgende studieplannen:
Master of Applied Sciences and Engineering: Applied Computer Science: Standaard traject (enkel aangeboden in het Engels)
Master in de ingenieurswetenschappen: computerwetenschappen: afstudeerrichting Artificiële Intelligentie
Master in de ingenieurswetenschappen: computerwetenschappen: afstudeerrichting Multimedia
Master in de ingenieurswetenschappen: computerwetenschappen: afstudeerrichting Software Languages and Software Engineering
Master in de ingenieurswetenschappen: computerwetenschappen: afstudeerrichting Data Management en Analytics