New

The latest blog posts to take off on Hacker News. 49,258 posts.

Any topic
Any year
29527

A New Angle on L2 Regularization

Deep neural networks have been shown to be vulnerable to the adversarial example phenomenon: all models tested so far can have their classifications dramatically altered by small image perturbations . The following predictions were for instance made by a state-of-the-art network trained to recognize celebrities :

29531

Lost in Math

Sabine Hossenfelder’s new book Lost in Math should be starting to appear in bookstores around now. It’s very good and you should get a copy. I hope that the book will receive a lot of a…