MetaInit: Initializing learning by learning to initialize

By Yann Dauphin, Sam Schoenholz. 2019

70 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Michal Drozdzal
2. Boris Knyazev
3. Emma Strubell
4. Sanket Vaibhav Mehta
5. Sarath Chandar
6. Darshan Patil
7. Adriana Romero-Soriano
8. Graham W. Taylor

Citations in the corpus

• An Empirical Investigation of the Role of Pre-training in Lifelong Learning (2021, methodology): “For this purpose, we employ the MetaInit algorithm (Dauphin and Schoenholz, 2019) to obtain a task-agnostic favorable initialization, which is independent of the specific task at hand.”
• Parameter Prediction for Unseen Deep Architectures (2021, methodology): “[117] Yann Dauphin and Samuel S Schoenholz.”
• Revisiting Few-sample BERT Fine-tuning (2020, unknown): “[6] Yann N Dauphin and Samuel Schoenholz.”