Neural Architecture Search and Google’s New AutoML Zero with Quoc Le
EPISODE 366
|
APRIL
16,
2020
Watch
Follow
Share
About this Episode
Today we're super excited to share our recent conversation with Quoc Le, a research scientist at Google, on the Brain team.
Quoc has been very busy recently with his work on Google's AutoML Zero, which details significant advances in automated machine learning that can "automatically discover complete machine learning algorithms just using basic mathematical operations as building blocks." Another major theme of this conversation is semi-supervised learning, discussing his work on the paper "Self-training with Noisy Student improves ImageNet classification." Finally, we discuss how his interest in sequence to sequence learning, and a chance encounter, led to the development of Meena, Google's recent multi-turn conversational chatbot.
This was a really fun conversation, so much so that we decided to release the video! April 16th at 12 pm PT, Quoc and Sam will premiere the video version of this interview on our Youtube page, and answer your questions in the chat. We'll see you there!
About the Guest
Quoc Le
Resources
- Paper: Self-training with Noisy Student improves ImageNet classification
- Video: Self-Training with Noisy Student (87.4% ImageNet Top-1 Accuracy!)
- Paper: Neural Architecture Search with Reinforcement Learning
- Paper: AutoML-Zero: Evolving Machine Learning Algorithms From Scratch
- Paper: Towards a Human-like Open-Domain Chatbot
- Paper: EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
