Daniel Fried

I will be starting as an assistant professor at the Language Technologies Institute in the School of Computer Science at Carnegie Mellon in Fall 2022. If you're a prospective student interested in working with me please apply through the SCS's Graduate Application this fall.

From August 2021--July 2022 I'll be a visiting researcher at Facebook AI Research in Seattle working with Mike Lewis and Luke Zettlemoyer.

I've recently graduated with a PhD from UC Berkeley where I was advised by Dan Klein in the NLP Group and the Berkeley AI Research Lab. Previously I did an M.Phil. at the Cambridge Computer Laboratory and a B.S. at the University of Arizona.

Links: Google ScholarCV [pdf], [html] Research Statement [pdf], [html]Teaching
Email: dfried@berkeley.edu Twitter: @dan_fried
Daniel Fried


Research Summary

I build language grounding systems that communicate with people to collaboratively solve real-world tasks. Language is a multi-agent system, and when people communicate with each other, they reason about the world and their conversational partners. Can our NLP systems do the same? One effective approach to grounded communication is pragmatics: viewing language as an action that produces effects on other agents and the world. This perspective has helped us build NLP systems that more successfully interact with people when generating (EMNLP 2021, NAACL 2018, NAACL 2019) and interpreting (NeurIPS 2018) language. Another effective approach to grounded communication is modularity: building neural modules that decompose a complex task (EMNLP 2021, NAACL 2021, ACL 2019).

I'm also broadly interested in NLP, in particular structured prediction (ACL 2020, TACL 2020, ACL 2018).

Talks on our work on: pragmatics and modularity.



Publications

Teaching

Course Materials

Teaching Experience