Overview

Deep learning means using neural networks with many layers to perform machine learning tasks, such as classifying images, detecting objects, predicting future actions, and interacting with the environment. Deep learning did not only revolutionize entire fields like computer vision or robotics, it also enables novel applications, like autonomous driving, in the first place. Hidden behind the buzzword generative artificial intelligence (GenAI), neural networks are used to generate realistic novel data, including text, images, videos, 3D models, and audio.

This lecture offers an overview on deep learning, reaching from fundamentals to cutting-edge applications.

Content

• Multi-layer perceptrons (MLPs) and training of deep neural networks
• Convolutional neural networks (CNNs) for computer vision
• Graph neural networks (GNNs) to process high-dimensional topological data
• Transformers and applications in natural language processing
• Generative AI (GenAI), including diffusion and latent space models
• Reinforcement learning for control and decision-making

The course combines lectures with hands-on exercises and includes paper presentations to expose students to current research. It is aimed at engineering students seeking a solid understanding of deep learning methods and their practical application in real-world systems.

Requirements

You should have experience with programming in Python or get familiar with its basics before the first exercise.