Research - YJ Choe

• CV

• Google Scholar

Research Interests

Broadly, I work on topics in statistics, machine learning, and natural language processing. In recent years, I've been excited about a somewhat eclectic set of research areas and topics:

• Game-theoretic statistics: sequential inference; anytime-validity; e-values and e-processes; confidence sequences; testing by betting; and evaluation of forecasters and black-box predictors;

• Science of large language models: causal representations; geometry of LLM embeddings; Transformers; (mechanistic) interpretability; and alignment.

Preprints

• Combining Evidence Across Filtrations Using Adjusters (arXiv, slides)
  Y. J. Choe, A. Ramdas
  arXiv, 2024
  Contributed talk at JSM 2024 (August); Invited talk at ICSDS 2024 (December)

• The Linear Representation Hypothesis and the Geometry of Large Language Models (arXiv, code, slides)
  K. Park, Y. J. Choe, V. Veitch
  arXiv, 2023
  Preliminary work presented (oral and poster) at the NeurIPS 2023 Workshop on Causal Representation Learning

Thesis

• Comparing Forecasters and Abstaining Classifiers (thesis, slides)
  Y. J. Choe
  Ph.D. Thesis, Carnegie Mellon University (2023)
  

Publications

Asterisks (*) denote equal contribution.

• Counterfactually Comparing Abstaining Classifiers (proc, arXiv, code, slides, poster)
  Y. J. Choe, A. Gangrade, A. Ramdas
  Advances in Neural Information Processing Systems (NeurIPS), 2023
  Preliminary work presented (oral and poster) at the ICML 2023 Workshop on Counterfactuals in Minds and Machines

• Comparing Sequential Forecasters (proc, arXiv, code, slides, poster)
  Y. J. Choe, A. Ramdas
  Operations Research, 2023
  Preliminary work presented at SAVI 2022 and JSM 2021
  Student Poster Award (Runner-up) w/ $10K Research Grant, Citadel Securities Inaugural PhD Summit

• KorNLI and KorSTS: New Benchmark Datasets for Korean Natural Language Understanding (proc, arXiv, data)
  J. Ham*, Y. J. Choe*, K. Park*, I. Choi, H. Soh
  Findings of the Association for Computational Linguistics: EMNLP, 2020

• word2word: A Collection of Bilingual Lexicons for 3,564 Language Pairs (proc, arXiv, code)
  Y. J. Choe*, K. Park*, D. Kim*
  Proceedings of the 12th Language Resources and Evaluation Conference (LREC), 2020

• Jejueo Datasets for Machine Translation and Speech Synthesis (proc, arXiv, code)
  K. Park, Y. J. Choe, J. Ham
  Proceedings of the 12th Language Resources and Evaluation Conference (LREC), 2020

• Predicting Drug–Target Interaction Using a Novel Graph Neural Network with 3D Structure-Embedded Graph Representation (proc, arXiv)
  J. Lim, S. Ryu, K. Park, Y. J. Choe, J. Ham, W. Y. Kim
  Journal of Chemical Information and Modeling, 2019

• A Neural Grammatical Error Correction System Built On Better Pre-training and Sequential Transfer Learning (proc, arXiv, code)
  Y. J. Choe*, J. Ham*, K. Park*, Y. Yoon*
  Proceedings of the 14th Workshop on Innovative Use of NLP for Building Educational Applications (BEA), 2019
  Runner-up, ACL Workshop Shared Task (Restricted & Low Resource Tracks)

• Discovery of Natural Language Concepts in Individual Units of CNNs (proc, arXiv, poster, code)
  S. Na, Y. J. Choe, D. Lee, G. Kim
  International Conference on Learning Representations (ICLR), 2019

• Local White Matter Architecture Defines Functional Brain Dynamics (proc, arXiv, slides)
  Y. J. Choe, S. Balakrishnan, A. Singh, J. M. Vettel, T. Verstynen
  IEEE International Conference on Systems, Man, and Cybernetics (SMC), 2018
  Franklin V. Taylor Memorial Award (for Best Paper and Oral Presentation)

Miscellaneous

• An Empirical Study of Invariant Risk Minimization (arXiv, code, slides)
  Y. J. Choe, J. Ham, K. Park
  ICML Workshop on Uncertainty and Robustness in Deep Learning, 2020

• Probabilistic Interpretations of Recurrent Neural Networks (report)
  Collaborators: J. Shin, N. Spencer
  Technical Report, 2017

• Learning Diverse Overcomplete Dictionaries via Determinantal Priors (abstract)
  M. Al-Shedivat, Y. J. Choe, N. Spencer, E. P. Xing
  ICML Workshop on Geometry in Machine Learning, 2016

• A Statistical Analysis of Neural Networks (report)
  Technical Report, 2016

• Sparse Additive Models with Shape Constraints (report, slides, code)
  Mentors: J. Lafferty, S. Chatterjee, M. Xu
  University of Chicago Computer Science REU, 2014