Learning from Noise
Chelsea discusses the innovative approach of using meta learning to enhance few shot and one shot learning. By intentionally corrupting demonstration labels with noise, the goal is to train systems to learn effectively from imperfect examples. This technique aims to improve adaptability across various tasks, showcasing the potential of learning how to learn.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Robotic Perception and Control with Chelsea Finn - #29
Related Questions