Humboldt-Universität zu Berlin - Mathematisch-Naturwissenschaftliche Fakultät - Institut für Informatik

Data-Efficient Approaches to Natural Language Processing

Abstract:

Natural Language Processing (NLP) aims to empower machines with the ability to understand and generate human language. A key milestone in recent years has been the development of large language models (LLMs) like ChatGPT, which can respond intelligently to a wide range of prompts. However, these models come with a significant limitation: they require vast amounts of data and computational resources to train effectively. This creates two major challenges: (1) many organizations lack the resources to train their own LLMs, and (2) many real-world applications remain unfeasible due to the scarcity of annotated data.

In this talk, I will explore how my research addresses these challenges by developing data-efficient NLP techniques. I will introduce two main areas of focus in my work: efficient large language modeling and practical information extraction from text. Additionally, I will showcase practical applications of these approaches, demonstrating how they can be used in real-world scenarios.

Die Veranstaltung findet digital per Zoom statt. Eine Zoom-Einladung finden Sie hier. (nur mit Informatik-Account)