AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Do You Know How to Interpolate Between Examples Well?
The challenge is how do you pick the most extreme examples and sample this space such that when you train your network, it's consistent at the points you train. And then by leaving some in your holdout set, you show that you're able to interpolate between examples well. What is the extent that you define in your data set, right? And you cover the space of what you will expect to likely see so that your system will adapt and perform well within that domain of the training data.