Generative AI for Text, Audio & Images



Next course


Further and more detailed information, including the schedule, can be found in the current course tables in the syllabus of the respective course, if the course is offered in the next sessions. The following text serves as information on what can be expected in terms of content in the course.

We will cover the basic concepts of prominent methods for generative deep learning before starting a deep dive on their application to text, audio and images. Since we will focus on applications and the usage of respective foundation models and toolkits, we strongly recommend you get familiar with deep learning ahead of this course. In detail, we will cover:

Theory: Prominent generative Deep Learning Methods
-Generative Pretrained Transformers (GPT), Fine-Tuning, RLHF, Instruction Learning, Zero-Shot Learning, In-Context Learning, Chain-of-Thought
-Generative Adversarial Networks (GAN)
-Variational Auto-Encoders (VAE)
-Style transfer

-Text: GPT Prompt Engineering
-Text to Speech, Voice Conversion
-Image generation and captioning

We will conclude the course with an outlook on risks, limitations, ethical and legal implications.