Adversarial Examples Are Not Bugs, They Are Features with Aleksander Madry - #369

episode

The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) not yet transcribed

▲ 0

Transcript

This episode hasn't been transcribed yet Upvotes decide what gets transcribed next — 0 so far. Sign in to upvote

Related

recent transcribed audio

Could sofa theory help you find your ideal partner? 4 min · 23 Sep 2026 · Do you really know?

Behind The Scenes At Vogue World Milano 28 min · 23 Sep 2026 · The Run-Through with Vogue

Is end-to-end encryption really secure? 4 min · 23 Sep 2026 · Do you really know?

Do tongue scrapers actually help with bad breath? 5 min · 23 Sep 2026 · Do you really know?

Lab Notes: Have we found dark matter? 10 min · 23 Sep 2026 · The Science Show

Is it cool to have no friends? 26 min · 23 Sep 2026 · Today in Focus

Report any issue

Source

Recorded27 Apr 2020

Publisherfeeds.megaphone.fm ↗

Handlerec:12496169

Show notes

Today we’re joined by Aleksander Madry, Faculty in the MIT EECS Department, to discuss his paper “Adversarial Examples Are Not Bugs, They Are Features.” In our conversation, we talk through what we expect these systems to do, vs what they’re actually doing, if we’re able to characterize these patterns, and what makes them compelling, and if the insights from the paper will help inform opinions on either side of the deep learning debate.

Ask this episode

“What did the council decide on item 42?” — answered with the speaker and the moment.

Sign in