🔥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

We (Deep learning theory team, RIKEN-AIP, Japan) are seeking a research scientist or postdoctoral researcher.

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

Joint work with Simon Shaolei Du Ken-ichi Kawarabayashi, Stefanie Jegelka, Mozhi Zhang Jingling Li

#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

Our work studies how the inductive bias in an architecture could impact its representations learned from noisy labels. We provide both empirical and theoretical support showing well-designed architectures can easily combat label noise even without robust training methods. 😃

VQ-GNN: A Universal Framework to Scale-up Graph Neural Networks using Vector Quantization abs: arxiv.org/abs/2110.14363

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

Chen out our #NeurIPS2021 paper "VQ-GNN: A Framework to Scale up GNNs using Vector Quantization" Poster session #7 7:30-9:00pm ET today (nips.cc/Conferences/20…) Paper link: arxiv.org/abs/2110.14363

We propose a universal and theoretically justified way to mini-batch train most Graph Neural Networks on large graphs, by approximating the remaining graph with vector quantization.

1/10: The summer break is the perfect time to share recent research from my lab. Our first story revolves around a fresh interpretation of diffusion-based generative modeling by my brilliant student Ying Fan. She proposed "diffusion models are solving a control problem".

🔥Check out our ICML Conference ICML23' work on training diffusion models with policy gradient for shortcuts, which is the first work to use RL for training diffusion models to our knowledge. Check out our Arxiv paper arxiv.org/abs/2301.13362 & an exciting follow-up work coming soon!

We proved Q* is hard 4 years ago 🤷‍♂️

Can large language models (LLMs) understand human beings (both individual and group wise) better than humans? Random thought @NeurIPS

I recently paid off my undergrad student loan from Bryn Mawr College, which had a 0% interest rate for nearly 11 years. I am deeply grateful for all the educational opportunities I've had over the years, and I still recall one of my CS professors joking that our first million was

⚡️ Excited to share that I am starting an AI+Education company called Eureka Labs. The announcement: --- We are Eureka Labs and we are building a new kind of school that is AI native. How can we approach an ideal experience for learning something new? For example, in the case

How to provably solve contextual goal-oriented problems (like controlling a robot via language cmd) using offline data? We present a very neat solution CODA, without the need of reward learning! Check out our poster tomorrow #NeurIPS2024 W Ballroom A-D #6206 (12/12, 11am-2pm)

Today we introduce an AI co-scientist system, designed to go beyond deep research tools to aid scientists in generating novel hypotheses & research strategies. Learn more, including how to join the Trusted Tester Program, at goo.gle/417wJrA

Our team at GDM is hiring! Apply here or share it with people who may be interested: deepmind.greenhouse.io/internal_job_b…

Our team is hiring! We are working on improving the social intelligence of the smartest models out there. Perks include: working with a diverse team of creative individuals (such Sian Gooding Jingling Li, naming those on X) One week left to apply 👇

🚨 I’m hosting a Student Researcher Google DeepMind! Join us on the Autonomous Assistants team (led by Edward Grefenstette ) to explore multi-agent communication—how agents learn to interact, coordinate, and solve tasks together. DM me for details!