Publications

Who should predict? Exact algorithms for learning to defer to humans
Hussein Mozannar, HL, Dennis Wei, Prasanna Sattigeri, Subhro Das, David Sontag.
AISTATS 2023.
[paper]

TabLLM: Few-shot Classification of Tabular Data with Large Language Models
Stefan Hegselmann, Alejandro Buendia, HL, Monica Agrawal, Xioayi Jiang, David Sontag.
AISTATS 2023.
[paper]

Large language models are zero-shot clinical information extractors
Monica Agrawal, Stefan Hegselmann, HL, Yoon Kim, David Sontag.
EMNLP 2022, oral presentation.
[paper] [dataset]

Training subset selection for weak supervision
HL, Aravindan Vijayaraghavan, David Sontag.
NeurIPS 2022.
[paper] [code]

Co-training improves prompt-based learning for large language models
HL, Monica Agrawal, Yoon Kim, David Sontag.
ICML 2022.
[paper] [code]

Leveraging time irreversibility with order-contrastive pre-training
Monica Agrawal*, HL*, Michael Offin, Lior Gazit, David Sontag.
AISTATS 2022, *equal contribution
[paper]

Graph cuts always find a global optimum for Potts models (with a catch)
HL, David Sontag, Aravindan Vijayaraghavan.
ICML 2021, long oral presentation.
[paper] [video]

Beyond perturbation stability: LP recovery guarantees for MAP inference on noisy stable instances
HL*, Aravind Reddy*, David Sontag, Aravindan Vijayaraghavan.
AISTATS 2021. *equal contribution
[paper]

Self-supervised self-supervision by combining deep learning and probabilistic logic
HL, Hoifung Poon.
AAAI 2021.
[paper]

Using statistics to automate stochastic optimization
HL, Pengchuan Zhang, Lin Xiao.
NeurIPS 2019.
[paper / code]

Understanding the role of momentum in stochastic gradient methods
Igor Gitman, HL, Pengchuan Zhang, Lin Xiao.
NeurIPS 2019.
[paper / code]

Block stability for MAP inference
HL, David Sontag, Aravindan Vijayaraghavan.
AISTATS 2019, long oral presentation.
[paper]

Optimality of approximate inference algorithms on stable instances
HL, David Sontag, Aravindan Vijayaraghavan.
AISTATS 2018.
[paper]

Preprints:

Statistical adaptive stochastic gradient methods
Pengchuan Zhang, HL, Qiang Liu, Lin Xiao.
arXiv, 2020.
[paper / code]