🔥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

Graph Foundation Models. Ultra Deluxe 💅. You know what to do. 3011 East Exhibition Hall Zhaocheng Zhu Bruno Ribeiro Jian Tang

Towards foundations… :) Following up on the math-related tweets. The math is double equivariance and the assumptions are in the double equivariance paper.

Can we design a GNN architecture that simplifies mitigating diverse and heterogeneous graph distribution shifts? Join us today at the NeurIPS poster session. I will present GraphMETRO, a mixture-of-experts GNN architecture. Stop by our poster session! 11 a.m. to 2 p.m. East

Our new workshop at ICLR 2025: Weight Space Learning: weight-space-learning.github.io Weights are data. We can learn from weights. Learning can outperform human-designed methods for optimization, interpretability, model merging, and more.

Slides of my presentation "Mathematical Foundations of Graph Foundation Models" yesterday at the AMS Session of the #JMM2025. The accompanying paper is coming soon. cs.purdue.edu/homes/ribeirob…

The most ubiquitous type of AI is (still) AI prediction. It powers everything from the TikTok algo (still there?) to the X feed . It drives trillions of market cap. If your company isn't using predictive AI - it may be a decade+ behind. Hema Raghavan Kumo

Great insights from Hema Raghavan and Konstantine Buhler 🚀 Predictive AI isn't just a buzzword—it’s the foundation of modern decision-making, from social feeds to enterprise operations. At Kumo, we're pushing the boundaries by making predictive AI more accessible and

🔬 Excited to announce our ICLR 2025 Workshop on *Weight Space Learning*! Applying deep learning to model weights could revolutionize how we develop, analyze & manipulate NNs. Submit by Feb 10 In-person @ Singapore + remote options ICLR 2026 #ICLR2025 weight-space-learning.github.io

🚨 Is your private data really safe in Federated Learning? Spoiler: Not always. 🚨 Attackers can reconstruct sensitive user data from model updates using gradient inversion attacks. We present CENSOR, a novel defense that breaks these attacks while keeping models utility. 🧵👇

7/ Joint work with amazing collaborators Siyuan Cheng, Guangyu Shen, Bruno Ribeiro, Shengwei An, Pin-Yu Chen , Xiangyu Zhang, Ninghui Li Purdue Computer Science IBM Research NDSS Symposium #NDSSsymposium2025

This solves an open problem in graph learning: How to create node representations for multiple order tasks? The solution is interesting since it demands defining a generalization of eigenvalues for GNN representations, where eigenvalues/eigenvectors are just a special case

ICLR will be in Brazil next year!

Bill Clinton invested $3B in 1990 for the first de novo human genome ~95%. In 2022, T2T achieved the first 100% assembly. Our AI assembler is the first of its kind - demonstrating this challenge can be solved w/ more data and compute. A call to industry: let's finish the job!