Modules

Prerequisites

Contents

• Artificial Intelligence: definitions, scope, history, limitations • Machine Learning vs. Deep Learning: formulations, data, losses, generalization; MNIST case study • NLP (Natural Language Processing) fundamentals: text as data, tokenization, representations • Word embeddings: motivation, distributional semantics, strengths and weaknesses • CNNs and RNNs: explanation, comparison and analysis • Transformers: self-attention intuition, components, impact • Language Models and LLMs: pretraining, scaling, evaluation, failure modes • Practical use of LMs: fine-tuning, prompting, RAG, reasoning behavior and constraints • Overview of multimodal models: VLMs, VLAs etc.

Learning outcomes

After successful completion of the module students understand core concepts of supervised and unsupervised machine learning with a focus on NLP (Natural Language Processing); can explain and implement representation-learning methods from word embeddings to Transformers; can critically compare assumptions, objectives, trade-offs, and failure modes; can discuss foundation model families (LLMs, VLMs, VLAs); can design basic experiments, evaluate models, and communicate results clearly.

Assessment

• written examination, 90 minutes

Remarks

• Study programme Artificial Intelligence M.Sc.: Compulsory elective module in complex „Learning and Reasoning” • Study programme Informatik M.Sc.: Compulsory elective module in complex „Angewandte und Technische Informatik” (Niveaustufe 400) • Study programme Mathematics M.Sc.: Compulsory elective module in complex „Applications” • Study programme Mathematical Data Science M.Sc.: Compulsory elective module in complex „Fundamentals of Data Science” • Study programme Micro- and Nanoelectronics M.Sc.: Compulsory elective module in complex „Applications” • Study programme Medizininformatik B.Sc.: Compulsory elective module in complex „Informatik”