Gaoussou Youssouf Kebe
University of Maryland, Baltimore County, Carnegie Mellon University
Papers
6
Total Citations
25
H-Index
3
About
Gaoussou Youssouf Kebe is a researcher at the forefront of grounded language acquisition, working at the intersection of natural language processing, computer vision, and robotics. His work addresses one of the most challenging problems in AI: enabling robots to learn language by connecting words to real-world percepts, such as visual and auditory data. Kebe has made significant contributions by developing practical cross-modal manifold alignment techniques that use triplet loss to create consistent, multi-modal embeddings for language-based concepts. He has also pioneered the use of deep acoustic representations for grounded language learning, moving beyond traditional text-based inputs to incorporate raw speech. His datasets, including a spoken language dataset for speech-based grounded learning and the GeSTICS corpus for gesture synthesis, have been instrumental in advancing research in human-robot interaction. With over 25 citations across his most-cited works, Kebe’s research has demonstrated the feasibility of sample-efficient, deployment-ready models for virtual and physical human-robot interaction. His work on using virtual reality for training data collection represents a novel approach to overcoming data scarcity in grounded language learning, making him a key figure in the quest for more intuitive, natural interfaces between humans and machines.
Research Focus
Key Achievements
Top Papers
- 1
- 2Towards Making Virtual Human-Robot Interaction a Reality6 citations · 2021
- 3
- 4
- 5
- 6Practical Cross-modal Manifold Alignment for Grounded Language2 citations · 2020