New paper: Evolving Normalization-Activation Layers.
— Hanxiao Liu (@Hanxiao_6) April 8, 2020
We use evolution to design new layers called EvoNorms, which outperform BatchNorm-ReLU on many tasks. A promising use of AutoML to discover fundamental ML building blocks.https://t.co/qFfqGzPO6G
Joint work with @DeepMind pic.twitter.com/A98YVNzrtg