AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
3min chapter
147. Large language models (LLMs) - interview with Dr Guy Emerson
Education Bookcast
How to Teach a Strategy to Poke Hole
A computer vision model can be used to predict what a picture will look like if it's changed in such a subtle way that humans cannot see the difference. The technique is called simple gradient descent approach and was developed by researchers at University of California, San Diego. Using this you can make very small changes to all the individual pixels right to make it slightly less likely to predict them they make the correct prediction. Each step just looks completely indistinguishable for a human because you're just making random what looks like random changes to on a pixel level yeah but leaving the kind of high level structure the same. A recent finding there's also the the images where you if you change like three pick the one or three
00:00
Transcript
Episode notes
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode