Alexander Dazhen Cai

About

Alex Dazhen Cai is curious about language(s). His interests have led him to a Ph.D. in Computer Science advised by Kianté Brantley and supported by a Kempner Graduate Fellowship, for whom and which he is very grateful. Alex completed his bachelor’s and master’s degrees at Harvard, during which he studied German, interned at DataStax and QuantCo, and authored a course textbook on reinforcement learning. He enjoys journaling, historical fiction, and running Dungeons and Dragons games in his spare time.

Research Focus

How does human language-learning differ from language-model training? Infants have access to a wealth of cognitive tools such as intention reading, joint attention, and cultural learning that allow them to robustly generalize from limited data. Alex hopes to borrow from sequential decision-making frameworks such as reinforcement learning and imitation learning to better model language acquisition, design more sample and compute-efficient training methods, and thereby unlock language models for low-resource languages.