HCDS at DLI2025 – Best Paper Award for African Language Research

27 August 2025, by Janis-Marie Paul

Photo: Janis-Marie Paul

The Deep Learning Indaba (DLI) is the largest gathering of the African AI community, bringing together researchers, students, and practitioners from across the continent each year. Its mission is to strengthen excellence in machine learning and AI, foster exchange across regions, and build sustainable structures for research and education.

Our Technical Lead Dr. Seid Muhie Yimam from the Hub of Computing and Data Science (HCDS), University of Hamburg, attended the Deep Learning Indaba 2025 – an inspiring opportunity to connect with leading AI and NLP researchers across Africa.

Best Paper Award for African Languages
We are proud to share that our joint paper
“The State of Large Language Models for African Languages: Progress and Challenges” received the Best Paper Award.

(Authors: Kedir Yassin Hussen, Walelign Tewabe Sewunetie, Abinew Ali Ayele, Sukairaj Hafiz Imam, Eyob Nigussie Alemu, Shamsuddeen Hassan Muhammad & Seid Muhie Yimam)

Key insights:

• Only 42 African languages are currently supported in popular LLMs and multilingual models
• Out of 23 actively used African scripts, only 3 are represented.
• Major challenges remain: data scarcity, lack of benchmarks, and high computational costs.

Presentation on AI for Conservation

In addition, we had the opportunity to present our paper  “Evaluating Deep Learning Models for African Wildlife Image Classification: From DenseNet to Vision Transformers”. The work compares CNNs and Vision Transformers for African wildlife image classification, highlighting the trade-off between accuracy and resource requirements for practical conservation deployment.

Towards Inclusive AI for African Contexts
Both contributions emphasize the importance of developing inclusive and practical AI solutions tailored to African contexts, helping to reduce global inequities in technology development.

Urunana — hand in hand.