AI Seminar: Nora Hollenstein
Interpreting Character Embeddings With Perceptual Representations: The Case of Shape, Sound, and Color
Character-level information is included in many NLP models, but evaluating the information encoded in character representations is an open issue. We leverage perceptual representations in the form of shape, sound, and color embeddings and perform a representational similarity analysis to evaluate their correlation with textual representations in five languages from PPMI, LSTM and transformer character models. We further develop a set of probing classifiers to intrinsically evaluate what phonological information is encoded in character embeddings.
Before joining the Center for Language Technology at the University of Copenhagen, Nora was a PhD candidate at DS3Lab at ETH Zurich working on cognitively-inspired natural language processing. She was also a lecturer at the Institute of Computational Liniguistics of the University of Zurich.
The focus of Nora's research lies in enhancing NLP applications with cognitive data such as eye-tracking and brain activity recordings. She is especially interested in multi-modal learning, learning from limited data, and the interpretability and cognitive plausibility of machine learning models.
This seminar is a part of the AI Seminar Series organised by SCIENCE AI Centre. The series highlights advances and challenges in research within Machine Learning, Data Science, and AI. Like the AI Centre itself, the seminar series has a broad scope, covering both new methodological contributions, ground-breaking applications, and impacts on society.