How early can we average Neural Networks?
Name
nasimov-unasimov-meng-eecs-2023-thesis.pdf
Description
Thesis PDF
Size
2.12 MB
Format
Adobe PDF
Checksum (MD5)
abf542f1c303626ae611e5f5b197e84c
Author(s)
Nasimov, Umarbek
Advisor(s)
Poggio, Tomaso
Date Issued
June 2023
Publisher
Massachusetts Institute of Technology
Abstract
There is a recurring observation in deep learning that neural networks can be combined simply with arithmetic averages over their parameters. This observation has led to many new research directions in model ensembling, meta-learning, federated learning, and optimization. We investigate the evolution of this phenomenon during the training trajectory of neural network models initialized from a common set of parameters (parent). Surprisingly, the benefit of averaging the parameters persists over long child trajectories from parent parameters with minimal training. Furthermore, we find that the parent can be merged with a single child with significant improvement in both training and test loss. Through analysis of the loss landscape, we find that the loss becomes sufficiently convex early on in training, and, as a consequence, models obtained by averaging multiple children often outperform any individual child.
MIT Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
