Language modeling via stochastic processes

RE Wang, E Durmus, N Goodman… - arxiv preprint arxiv …, 2022 - arxiv.org
Modern language models can generate high-quality short texts. However, they often
meander or are incoherent when generating longer texts. These issues arise from the next …

Contrastive estimation reveals topic posterior information to linear models

C Tosh, A Krishnamurthy, D Hsu - Journal of Machine Learning Research, 2021 - jmlr.org
Contrastive learning is an approach to representation learning that utilizes naturally
occurring similar and dissimilar pairs of data points to find useful embeddings of data. In the …

Leveraging time irreversibility with order-contrastive pre-training

MN Agrawal, H Lang, M Offin… - International …, 2022 - proceedings.mlr.press
Label-scarce, high-dimensional domains such as healthcare present a challenge for
modern machine learning techniques. To overcome the difficulties posed by a lack of …

Towards scalable structured data from clinical text

M Agrawal - 2023 - dspace.mit.edu
The adoption of electronic health records (EHRs) presents an incredible opportunity to
improve medicine both at the point-of-care and through retrospective research …

On the Sequence Evaluation based on Stochastic Processes

T Zhang, Z Lin, Z Sheng, C Jiang, D Kang - arxiv preprint arxiv …, 2024 - arxiv.org
Modeling and analyzing long sequences of text is an essential task for Natural Language
Processing. Success in capturing long text dynamics using neural language models will …

TRAPL: Transformer-Based Patch Learning for Enhancing Semantic Representations Using Aggregated Features to Estimate Patch-Class Distribution

SR Jyhne, PA Andersen, I Oveland… - … Conference on Innovative …, 2024 - Springer
We introduce TRAPL, a Transformer-based Patch Learning technique that enhances
semantic representations in segmentation models. TRAPL leverages aggregated features …