Course details


Advanced NLP

WS 2023 Dr. Elia Bruni OFFLINE
B.Sc modules:
CS-BWP-AI - Artificial Intelligence
CS-BWP-CL - (Computational) Linguistics
CS-BWP-MCS - Methods of Cognitive Science
CS-BWP-NI - Neuroinformatics
KOGW-WPM-CL - Computational Linguistics
KOGW-WPM-KI - Artificial Intelligence
KOGW-WPM-NI - Neuroinformatics
M.Sc modules:
CC-MWP-AI - Artificial Intelligence
CC-MWP-CL - Computational Linguistics
CC-MWP-NI - Neuroinformatics
CS-MWP-AI - Artificial Intelligence
CS-MWP-CL - (Computational) Linguistics
CS-MWP-NI - Neuroinformatics

CS-BW - Bachelor elective course
CS-MW - Master elective course

The course will provide a historical perspective on deep learning for natural language processing (NLP) and will address recent topics such as Transformers (e.g., BERT and GPT), attention-based models and recent models for dialogue. In addition, we will discuss language acquisition, the cognitive plausibility of AI models, and the extraction of semantic structure from raw text. We will take a look at the current revival of linguistic structure in the deep learning community, either through the analysis of attention patterns in Transformers (according to which linguistic structure is a 'by-product' of neural attention) or through diagnostic classifiers. We will go through a bit of theory in the first part of every lecture, and proceed with a discussion of recent literature in the second part, with an active role for students which will introduce papers on the collective reading list and work in groups on short practicals. Course objectives: Students will obtain knowledge about the historical and current trends in deep learning-based NLP. They will be able to take a critical look at current literature and will have a rather advanced understanding of the challenges, opportunities and pitfalls of deep learning applied to language. Furthermore, they will have obtained practical knowledge about how to instantiate some of the latest NLP models. Prerequisites: Basic programming; Deep Learning for NLP or other deep learning background.