State of Technology for African Languages Lecture

Large language models (LLMs) have achieved impressive results in a wide range of natural language applications. However, they often struggle to recognize low-resource languages, in particular African languages, which are not well represented in large training corpora.

In this talk, Dr. Happy Buzaaba will highlight the need to create technology for low-resource African languages and our contributions/current work towards increasing the representation of African languages in NLP and LLMs/AI research.

About Dr. Buzaaba:
I am a postdoctoral research associate at the Princeton Center for Digital Humanities, and my work focuses on multilingual NLP for low-resource languages especially African languages. Before that, I was a postdoctoral researcher at the RIKEN Centre for Advanced Intelligence Project, in the Approximate Bayesian Inference Team where I spent time investigating the use of natural-gradient Bayesian methods to improve uncertainty estimation in PLMs.

I completed my Ph.D. in the Knowledge & Data Engineering Lab at the University of Tsukuba in Japan. During my Ph.D, I was lucky to be advised by Prof. Toshiyuki Amagasa on making use of machine learning and computational linguistics to reason about language and knowledge. I focused on question answering over knowledge base systems.