🔥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

cc @bowlinearl Ronald E Robertson Jon Green Damian Ruck

Damian Ruck Ronald E Robertson Jon Green Katherine Ognyanova (@ognyanova.bsky.social) @bowlinearl 🇺🇦 [email protected] Damian, I posted about it here. techpolicy.press/strong-republi…

Great thread. Point one is especially insightful.

Did religion facilitate modern growth? Or was it a stumbling block? Using data for 912 cities and 500,000 individuals born throughout Europe during the past 700 years, we find evidence for the latter. New CEPR wp w/ Lars Harhoff Andersen: tinyurl.com/yytkbavd More details below

O…M…G

Google open-sourced Wikipedia Image-Text Dataset. - 37.5 million entity-rich image-text examples - 11.5 million unique images - across 108 languages Find it here: github.com/google-researc…

We don’t talk enough about this

"Inequality between identity groups and social unrest" w/ C. Houle, Damian Ruck, Alex Bentley bit.ly/3x780X8 Royal Society Publishing

This should be a pretty distressing figure for many researchers using Twitter...

Today in collaboration with Massachusetts Institute of Technology (MIT) and Cornell University we announce a new SOTA for unsupervised Semantic Segmentation. “STEGO” discovers the world’s objects and classifies every pixel of input images: aka.ms/stego-news

Check out the COVID States analysis of this odd moment in the pandemic, where there (seems to be) a strong _positive_ relationship between vaccination rates and cases (but, not deaths) at the state level. covidstates.org/blog/an-odd-pa…

Awareness of “woke” and “cancel culture” has increased between 2020 and 2022 in the UK. ipsos.com/en-uk/woke-can…

Today I got a call by someone asking me to participate on a survey. I could feel the scam and when I asked how they got my number, they replied "that's none of your business, motherfucker". No surprise that telephone response rates are still falling. pewresearch.org/fact-tank/2019…

Adversarial examples, originally observed ~2014, were fascinating cases showing deep nets can be super sensitive to small input perturbations. Since then a lot has happened! Did we solve this failure mode of AI models or have we hit a wall? A thread 🧵

Queen Elizabeth reigned over both Alan Turing and the GPU.

"Real Attackers Don't Compute Gradients" – is the opening of the paper (just accepted to SaTML Conference!) that I loved the most writing. So glad of having found such a great and passionate team! Fabio Pierazzi Hyrum Anderson Savino Dambra Kevin Alejandro Roundy & David Freeman

Today I met machine learning’s number one enemy #spuriouscorrelations

Writing this book and finally seeing it appear on my bookshelf has been a very joyful journey. Truly appreciate Ben Dickson's kind words for highlighting our book on "Adversarial Robustness for Machine Learning", co-authored with Cho-Jui Hsieh UCLA CS Department. Also kudos to IBM Research

I was thrilled to be involved in this study, just out in nature. Led by the awesome Ronald E Robertson, it sheds new light on why people engage with unreliable and partisan news when they use Google Search. Check out the thread below.

Really interesting talk on the rapidly emerging field of LLM-ops. youtu.be/Fquj2u7ay40