
23 - Mechanistic Anomaly Detection with Mark Xu
AXRP - the AI X-risk Research Podcast
Distribution vs. Out of Distribution Anomaly Detection
I think the main question is like whether or not it's okay to get tripped up by seeing new rare things in distribution. I'm not even sure what the distinguishing features of those are. But there's a naive thing that I want to do, which is like take a training set, whitelist all the mechanisms that happen during training set and flag everything else as anomalous. And then if something is anomalous, we can like inspect it carefully and check whether ornot it's actually anomalous. Right. It's not anomalous. We added to the whitlist. Or if it's like actually not anomalous,we added to theWhitelist. The hope is that
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.