3min chapter

Machine Learning Street Talk (MLST) cover image

#92 - SARA HOOKER - Fairness, Interpretability, Language Models

Machine Learning Street Talk (MLST)

CHAPTER

How to Identify the Junk in a Neural Network Image Classifier

This method is actually very interesting way of doing that. So within common crawl you have a lot of junk from the internet like HTML code or you have a Reddit like you know kind of gibberish like they're nonsensical so often these are removed right now using these rule based approaches. This paper does an opposite it says let's optimize explicitly for this by making it part of the process to distill what is important and what are the important bits within the image that are important for end performance. It's also getting to the heart of what does the model actually need to perform well.

00:00

Get the Snipd
podcast app

Unlock the knowledge in podcasts with the podcast player of the future.
App store bannerPlay store banner

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

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