Machine-learning-based image classifiers are vulnerable to "adversarial preturbations" where small, seemingly innocuous modifications to images (including very trivial ones) can totally confound them. (more…)
Popular Posts
-
Looking for something to illustrate a post about crunch-time in game development, I ran into this video depicting many forms of footwear (...
-
The guest this week on my Cool Tools show is Madeline Ashby . Madeline is a science fiction writer and futurist living in Toronto. Her mos...
-
The Sound Archive posted a 1930s-era recording of a conversation in a British pharmacy . The received-pronunciation chatter isn't quite ...
-
Dropbox syncs files between computers, but who wants that? The New Dropbox , announced this week, has all sorts of wonderful features to or...
-
Writing in Marker, David Gauvey Herbert gives us an extended-play version of China's legendary bank-robber, Ren Xiaofeng, a bank officia...
-
While some who saw this variable message sign in Houston though it had been hacked, it's actually a roadside art installation by Brookly...
-
Christopher Ferguson, an off-duty cop in Algood, Tenn., going 20 miles over the speed limit, will not be inconvenienced after ramming into ...
-
https://vimeo.com/71952791 What happens if you allow a group of onlookers to do anything they want to you for six hours? Marina Abramovich ...
-
Beware the rogue .wav file. Two reports published in the last few months indicate that authors of malware programs are using an interestin...
-
"Come and play with us, Danny... for ever, and ever, and ever." The bigscreen adaptation of Doctor Sleep , Stephen King's 201...
