Taylor Berg-Kirkpatrick

Natural Language Processing and Machine Learning

Unsupervised Learning
Developing models and methods that can learn to understand aspects of human language without relying on hand-labeled data

Structured Prediction
Building inference procedures that can efficiently and accurately produce complex outputs, including both structured linguistic annotations (like parse trees) and natural language utterances (like summaries)

Analysis of Human Artifacts
Using automated methods to analyze and decipher structured human artifacts, including text but also sources like piano music, images of early modern books (e.g., First Folio of Shakespeare), and cuneiform tablets