Interpretable generative AI, language in humans, animals, and machines
Gasper Begus, Professor
Linguistics
Check back for status
This project looks at language in deep neural networks, animals (whales, elephants, spiders), and humans. We model spoken language from raw audio using deep generative neural networks (GANs). We use audio, neural, and behavioral data in spoken language to better understand and interpret deep learning models.
Role: The undergraduate student will help with:
(i) coding the models in Pytorch
(ii) audio signal processing
(iii) helping with data analysis (acoustic and statistical analysis in Praat, Librosa, R), interpreting the results
Qualifications: Coding knowledge in PyTorch (or Tensorflow) required, knowledge of machine learning theory. Students with interest in CS, cognitive science, linguistics, marine biology, and signal processing are very welcome to apply.
Hours: to be negotiated
Related website: https://www.gasperbegus.com
Related website: https://docs.google.com/document/d/1xrRXGssVcetcKnZ0wOORXgKE8WGaxahq3xDNF9juF0o/edit?usp=sharing