Gašper Beguš is an Assistant Professor of Linguistics at the Department of Linguistics at UC Berkeley. Gašper obtained his undergraduate degree at the University of Ljubljana and his MA and Ph.D. at Harvard University. At Harvard, he served as a Resident Tutor and Sophomore Advising Coordinator, and was awarded the Certificate of Teaching Excellence award, Mind Brain Behavior Graduate Student award, Harvard Merit/Term-Time award, and was nominated for the Star Family Prize in Excellence in Advising. The focal question of his research is how humans acquire speech and how we can computationally simulate this acquisition. He combines experimental, statistical, and computational models to address this question. He is currently exploring how the well-understood dependencies in speech data can help us understand what and how neural networks learn and how they encode learning representations. His work appeared in the Journal of Linguistics, Journal of the Acoustical Society of America, Journal of the American Oriental Society.
Research projects: Prof. Beguš research focuses on developing deep learning models for speech data. More specifically, he trains models to learn representations of spoken words from raw audio outputs. He combines machine learning with behavioral experiments and statistical models to better understand how neutral networs learn internal representations in speech and how humans learn to speak. He has worked and published on sound systems of various language families such as Indo-European, Caucasian, and Austronesian languages.