Laure Thompson

Cornell University

214 Gates Hall

laurejt [at] cs [dot] cornell [dot] edu

Notice: I am currently on the academic job market!

I am a final-year PhD candidate in Computer Science at Cornell University where I am advised by David Mimno. My research bridges machine learning and natural language processing with humanistic scholarship. My work focuses on understanding what computational models learn and how we can intentionally change what they learn. I develop methods for quantitatively measuring how information is encoded within trained models' latent vector spaces and how intentional modification of the underlying training data affects what these models learn. Since my work is centered on humanities applications, I work with a wide range of cultural heritage collections: from texts of science fiction novels and medieval manuscripts to images of avant-garde journals and magical gems.

I am a recipient of a NSF Graduate Research Fellowship and Cornell University Fellowship.
My work has received a best paper award at COLING and an honorable mention at EMNLP.

Publications

Computational Cut-Ups: The Influence of Dada.
Laure Thompson and David Mimno.
The Journal of Modern Periodical Studies Vol.8, No.2, 2018, pp. 179-195.
[JSTOR] [Project MUSE] [preprint]

Authorless Topic Models: Biasing Models Away from Known Structure.
Laure Thompson and David Mimno.
Conference on Computational Linguistics (COLING), 2018.
Best Paper Award: Best NLP Engineering Experiment.
[pdf] [repo]

The Strange Geometry of Skip-Gram with Negative Sampling.
David Mimno and Laure Thompson.
Conference on Empirical Methods in Natural Language Processing (EMNLP), 2017.
Outstanding Paper (Best Paper Honorable Mention).
[pdf]

Quantifying the Effects of Text Duplication on Semantic Models.
Alexandra Schofield, Laure Thompson, and David Mimno.
Conference on Empirical Methods in Natural Language Processing (EMNLP), 2017.
[pdf]

Vale: Verifying High-Performance Crytographic Assembly Code.
Barry Bond, Chris Hawblitzel, Manos Kapritsos, Rustan Leino, Jay Lorch, Bryan Parno,
Ashay Rane, Srinath Setty, and Laure Thompson.
USENIX Security Symposium, 2017.
Distinguished Paper Award.
[pdf]

A Coalgebraic Decision Procedure for NetKAT.
Nate Foster, Dexter Kozen, Matthew Milano, Alexandra Silva, and Laure Thompson.
ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages (POPL), 2015.
[pdf]

Talks

Understanding and Directing What Models Learn.
Machine Learning and Friends Lunch (MLFL) series, University of Massachusetts Amherst, 2019.

Analysis of Visual Corpora with Deep Learning.
Chair and panelist.
The Association for Computers and the Humanities (ACH) Conference, 2019.

Understanding and Directing the Learning of Latent Vector Spaces.
Graduate student colloquium, Rutgers University, 2019.

Medieval MALLET Mishaps: Topic Modeling Difficult Corpora.
Joint talk with David Mimno and Anna Fore Waymack.
54th International Congress on Medieval Studies

Authorless Topic Models: Biasing Models Away from Known Structure.
Poster presentation.
Ninth Annual Conference on New Directions in Analyzing Text as Data (TADA), 2018.

Authorless Topic Models: Biasing Models Away from Known Structure.
Conference on Computational Linguistics (COLING), 2018.

Data Analysis for the Humanities.
Cornell Summer Graduate Fellowship program, 2018.

Bilingual Topic Modeling in the Patrologia Graeca
Joint talk with David Mimno.
Future Philologies: Digital Directions in Ancient World Text, 2018.

The Strange Geometry of Skip-Gram with Negative Sampling: A story of geometric observations.
Conference on Empirical Methods in Natural Language Processing (EMNLP), 2017.

Measuring Oracular Authenticity: Distinguishing the Historical from the Legendary in the Oracles of Delphi.
LAWDNY Digital Antiquity Workshop 2016 at ISAW.

A Coalgebraic Decision Procedure for NetKAT.
ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages (POPL), 2015.

Teaching Assistantships

Cornell University

University of Washington

Last updated November 2019.

I stole this page from Franzi Roesner.