Reporting Uncertainty
Rishabh discusses the complexities of calculating variance when metrics become less straightforward, emphasizing the importance of simulating distributions. He highlights the potential for adopting established tools from computer science to improve performance reporting and suggests that there may be more effective methods for conveying uncertainty, hinting at the need for theoretical guarantees in these approaches.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Deep Reinforcement Learning at the Edge of the Statistical Precipice with Rishabh Agarwal - #559
Related Questions