Deep Learning for Text Analytics - Detailseite

The module Deep Learning for Text Analytics introduces students to recent developments in the scope of deep learning and natural language processing. We first examine different forms of artificial neural networks, which are the backbone of modern AI-Systems. Special emphasis is given to the analysis of sequential data like time-series. Next, textual data is introduced as a special form of sequential data. We elaborate on seminal approaches and contemporary practices to process textual data and the corresponding applications. Frameworks and practices to use advanced (deep) machine learning technology and deploy corresponding solutions are of critical importance and will be elaborated in tutorial sessions.

The topics covered in the module include but are not limited to:

• Fundamentals of artificial neural networks
• Recurrent and convolutional neural networks for sequential data processing
• Fundamentals of natural language processing (NLP)
• Word embedding and language models
• NLP transfer learning
• Recent developments in NLP and AI

The module is designed as a follow-up to the module Business Analytics and Data Science. We recommend students to first complete that module. More specifically, it is strongly recommended to join Deep Learning for Text Analytics with a solid understanding of machine learning practices and algorithms. Experience with Python programming is also expected since we use the Python programming language in tutorials.

Students who have passed the exam 707929 Advanced Data Anaytics for Management Support are not allowed to take the module 707932 Deep Learning for Text Analytics!

A Zhang, ZC Lipton, M Li, AJ Smola (2020) Dive into Deep Learning, interactive deep learning book with code. https://d2l.ai/

Portfolio exam:

Element 1: Feedforward neural network for time series forecasting

Type: Programming task

Submission format: Forecasts for a test data set (e.g., via www.Kaggle.com)

Deadline: 10 Mai 2024

Weight 10%

Element 2: Recurrent neural network for time series forecasting

Type: Programming task

Submission format: Python codes

Deadline: 31 Mai 2024

Weight 20%

Element 3: Natural language processing

Type: Empirical study

Submission format: Jupyter Notebook

Deadline: 31 August 2024

Weight 35%

Element 4: Presentation/Defense of the Element 3 solution

Type: Oral exam

Submission format: n.A.

Deadline: September 2024

Weight 35%

Exam registration via AGNES: 16.04. until 09.05.2024 / deregistration until 09.05.2024.

Masterstudium