AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Using Contrastive Learning to Improve Robotic Learning
In contrastive learning, what you do as you have you want to learn what's in an image. But how'd you learn it if nobody tells you what's in an images? Well here's the idea. Imagine you just down lit two images, and you don't know what's in them. You have no idea what's in them - but they're different from one another. And that's really where the signal comes from. Now you train your nel network to understand that these two crops of that same first image are similar. We can do the same thing in reinforcement learning from images. So you apply the same idea, what the robot is seeing now first is what it's