Meet the recipients of the 2024 ACM A.M. Turing Award, Andrew G. Barto and Richard S. Sutton! They are recognized for developing the conceptual and algorithmic foundations of reinforcement learning. Please join us in congratulating the two recipients! bit.ly/4hpdsbD

Why are diffusion model creative? #CPAL2025 invited speaker Surya Ganguli gives a compelling theory that it is creative because it mixes and matches local patches from many different training images.

#CPAL2025 invited speaker Yuandong Tian shows that attention map first goes *sparser* than gets *denser* as we train a nonlinear attention mechanism using gradient flow — and the phenomena show up in the experiments too.

Okay. So I am going to Singapore for #ICLR2025 to present arxiv.org/abs/2402.05864. Interestingly, this will be my first in-person ICLR since 2015, when it collocated with #AISTATS in San Diego. A lot has changed in a decade. ICLR'15 program: iclr.cc/archive/www/20…

Curious about the state of watermarking and its role in AI safety? I will be at #ICLR Poster 237 (Hall 3) from 3pm this afternoon. Come and have a chat!

It’s very fitting that the watermarking workshop at #ICLR starts with Scott Aaronson. Interesting program for the rest of the day too.

The secret is revealed! The six ways images were watermarked in the blackbox track of the #NeurIPS2024 competition for watermark removal attacks. Guess what the most successful attacks were?

An upcoming workshop on "Inference Optimization for Generative AI" organized by a few awesome people I worked with at AWS. If you are going to #KDD2025, don't miss the workshop! Link to CfP: magic-town-d89.notion.site/KDD-2025-Works…

Think PII scrubbing ensures privacy? 🤔Think again‼️ In our paper, for the first time on unstructured text, we show that you can re-identify over 70% of private information *after* scrubbing! It’s time to move beyond surface-level anonymization. #Privacy #NLProc 🔗🧵

🚀 Excited to share the most inspiring work I’ve been part of this year: "Learning to Reason without External Rewards" TL;DR: We show that LLMs can learn complex reasoning without access to ground-truth answers, simply by optimizing their own internal sense of confidence. 1/n

new paper from our work at Meta! **GPT-style language models memorize 3.6 bits per param** we compute capacity by measuring total bits memorized, using some theory from Shannon (1953) shockingly, the memorization-datasize curves look like this: ___________ / / (🧵)

I am offering a course on *Safety in Generative AI* at UCSD next fall. I'd like to focus on the most important problems today. Mighty #AcademicTwitter: what are some topics that you think I must cover?