Transition to Linearity of Wide Neural Networks is an Emerging Property of Assembling Weak Models

10 March 2022

Papers citing "Transition to Linearity of Wide Neural Networks is an Emerging Property of Assembling Weak Models"

4 / 4 papers shown

Title
On the Role of Initialization on the Implicit Bias in Deep Linear Networks Oria Gruber H. Avron AI4CE 13 0 0 04 Feb 2024
Catapults in SGD: spikes in the training loss and their impact on generalization through feature learning Libin Zhu Chaoyue Liu Adityanarayanan Radhakrishnan M. Belkin 124 15 0 07 Jun 2023
On Emergence of Clean-Priority Learning in Early Stopped Neural Networks Chaoyue Liu Amirhesam Abedsoltan M. Belkin NoLa 61 5 0 05 Jun 2023
Transition to Linearity of General Neural Networks with Directed Acyclic Graph Architecture Libin Zhu Chaoyue Liu M. Belkin GNN AI4CE 60 4 0 24 May 2022