🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Our ICLR 20 (spotlight) paper: What can neural networks reason about? Why Graph Neural Networks generalize in many reasoning tasks (e.g. physics, visual questions, graphs)? Hint: the underlying reasoning procedure resembles dynamic programming. openreview.net/forum?id=rJxbJ…

I'll present work on cross-lingual text classification between orthographically similar languages at #AAAI2020. Visit our poster (4415) Tuesday night! Joint work with Yoshinari Fujinuma and Jordan Boyd-Graber arxiv.org/abs/1812.09617

Day 3 of posting ACL videos! Here's Yoshinari Fujinuma and Mozhi Zhang talking about why overfitting multilingual embeddings can hurt bilingual lexicon induction (the typical eval metric) but help downstream tasks. Video: youtube.com/watch?v=yVN47w… Paper: users.umiacs.umd.edu/~jbg//docs/202…

New #acl2020nlp paper - "Toward Gender-Inclusive Coreference Resolution." Paper: ycao95.umiacs.io/docs/1590.pdf. Talk and discussion: virtual.acl2020.org/paper_main.418…

How do neural networks extrapolate, i.e., predict outside the training distribution? We study MLPs and Graph Neural Networks trained by gradient descent, and show how a good representation and architecture can help extrapolation. arxiv.org/abs/2009.11848

Associate professor Jordan Boyd-Graber is presenting two papers next week .emnlp2020 on how to quickly train AI systems to sort through medical data: go.umd.edu/SNk

How do we use feedback from bilingual speakers to improve cross-lingual word embeddings? Join Gather Session 4A .emnlp2020 to discuss 'Interactive Refinement of Cross-Lingual Word Embeddings' with Michelle Yuan Mozhi Zhang & Jordan Boyd-Graber #EMNLP2020 aclweb.org/anthology/2020…

Continuing the experiment of reading papers out loud based on suggestion of blind student. This is "Interactive Refinement of Cross-Lingual Word Embeddings". Last one I had recorded, let me know if this worth keeping up. youtu.be/UqXpBm8pskM

1/3 GNNs are implicitly accelerated by skip connections and more depth. We also show gradient descent training linearized GNNs converge to global minima. arxiv.org/abs/2105.04550 w/ Kenji Kawaguchi, Mozhi Zhang Stefanie Jegelka

#Neurips2021 How does a Neural Network's Architecture Impact its Robustness to Noisy Labels? arxiv.org/abs/2012.12896 (Arxiv version will be updated soon) Joint work with Mozhi Zhang Keyulu Xu John P Dickerson, and Jimmy Ba

How does a neural network's architecture impact its robustness to noisy labels? Visit us at #Neurips2021 to learn more! When 🗓: Wed Dec 08, 7:30 - 9:00 PM (EST) Where📍: Slot B1 at eventhosts.gather.town/app/jS4PEO6rvP… UMD Department of Computer Science Joint work with Mozhi Zhang Keyulu Xu John P Dickerson Jimmy Ba