top of page

Generative AI (infGAI-01a)

Abstract

​

This lecture is dedicated to developing an understanding of generative AI models with a focus on visual computing. The lecture will introduce the diverse applications of generative AI, ranging from generating new data samples and using generative architectures for data augmentation to highlighting how these technologies are revolutionizing multiple industries. By the end of the lecture, attendees will gain an understanding of generative AI, equipped with the knowledge to critically assess its applications. This lecture is designed to cater to a wide range of students, from BA and MSc to students of other courses.


Learning Objectives

​

Students will be able to...

  • explain basic concepts of machine learning (optimizers, losses, backpropagation, etc.)

  • understand foundational principles of generative models, distinguishing them from traditional discriminative models in machine learning

  • implement various generative architectures, including Autoencoders, Generative Adversarial Networks (GANs), Diffusion Models and Transformers

  • describe their underlying mechanisms, strengths, and limitations

  • understand the difference between predictive and generative modeling and advanced concepts, such as synthetic data generation and various data augmentation strategies

  • discuss the technology's capabilities and limitations

  • to work in teams and to independently work on ML tasks

 
Course Content

​

  • Introduction to Deep Learning (Simple Architectures, Losses, Optimizers, Backpropagation)

  • Concepts of Generative Modeling

  • Convolutional Neural Networks

  • Autoencoders and variational autoencoders

  • Image-based predictive and discriminative network architectures (i.e., image-based tasks)

  • Transformers for image-based task

  • Generative Adversarial Networks

  • Domain Adaptation and Style Transfer

  • Diffusion Models

  • Neural Radiance Fields (NeRFs)

  • Synthetic Data Generation (with focus on images)

  • Data Augmentation Strategies

  • Transfer Learning and Finetune Training

 

Further Requirements

​

  • Basic knowledge about statistics, linear algebra, and especially differential calculus.

  • Familiarity with the programming language Phython.

  • All lecture slides and course material will be in English.

 

Exam


Written exam (100 min.). It is required to actively work on the exercises (homework) to be allowed to take the exam. The exam will be offered in the 2 examination time slots following the course.


Teaching and Learning Methods

 

Learning materials will be provided in the form of presentation slides. Primary lecture media is projected slide presentation. Occasionally complemented with drafts on board/white board. Concepts are introduced in the lectures with the help of examples and specific application tasks. In the exercise the knowledge is deepened and applied - guided by weekly homework assignments.


Literature

​

  • Deep Learning, Ian Goodfellow and Yoshua Bengio and Aaron Courville, 2016, MIT Press

  • Generative AI - Teaching Machines to Paint, Write, Compose and Play, David Foster, 2019, O'Reily

bottom of page