Modulcode:	infDLfNLP-01a
Englische Bezeichnung:	Deep Learning for Natural Language Processing
Modulverantwortliche(r):	Prof. Dr. Ralf Krestel
Turnus:	unregelmäßig (SS22 SS24 WS25/26)
Präsenzzeiten:	2V 2Ü
ECTS:	6
Workload:	30 Std. Vorlesung, 30 Std. Präsenzübung, 150 Std. Selbststudium
Dauer:	ein Semester
Modulkategorien:	MSc-Inf-WP (MSc Inf (21)) 2F-MEd-Inf-WP (MEd-Hdl Inf (21)) 2F-MA-Inf-WP (2F-MA Inf (21)) MSc-WInf-WP-WInf (MSc WInf (21)) WI (MSc Inf (15)) WWi (MSc WInf (15))
Lehrsprache:	Englisch
Voraussetzungen:	Inf-NNDL

Kurzfassung:

Deep learning approaches have surpassed many traditional machine learning methods in recent years in a variety of tasks. Besides for computer vision, this is particularly the case for natural language processing (NLP) and text mining. In this module, we give an introduction to basic concepts of deep learning for natural language processing. To this end, we present NLP tasks and deep learning architecutes used to solve these tasks. We have a detailed look at word embeddings and recurrent neural networks, as well as advanced, state-of-the-art transformer models and generative deep learning approaches.

The module comprises lecture and exercise parts to apply and help deepen the understanding of the presented methods. Further, homework assignments containing both theoretical and practical tasks prepare for the exam.

Lernziele:

Students will be able to...

• explain different network architectures for natural language processing tasks
• identify application areas of deep learning for natural language processing
• select suitable network architectures
• explain the functionality of different components of neural networks
• apply deep learning methods in Python
• design, implement, and evaluate their own text application
• understand the theoretical background, e.g., the backpropagation algorithmn
• recognize the limits and boundaries of deep learning
• discuss recent developments and have an overview of current research
• estimate societal impact of deep learning approaches

Lehrinhalte:

• Introduction
• Refreshing the Basics
• Word Embeddings
• Document Embeddings
• Knowledge Graph Embeddings
• Recurrent Neural Networks
• LSTM and GRU
• CNN for Text
• Practical Considerations
• Sequence-to-Sequence models
• Transformer and Attention
• Neural Topic Models
• Neural Information Retrieval
• GAN and VAE
• Optimization and Evaluation

Weitere Voraussetzungen:

• Basic knowledge about statistics, linear algebra, and especially differential calculus
• Familiarity with the programming language Phython

Prüfungsleistung:

• The final exam will be in form of a written exam.
• Solving the homework assignments are prerequisites for participation in the exam.

Lehr- und Lernmethoden:

Learning materials will be provided in form of presentation slides. Primary lecture media is projected slide presentation. Occasionally complemented with drafts on board/white board.

Verwendbarkeit:

Literatur:

• Deep Lerning with Python by Francois Chollet

Verweise:

Kommentar: