Search Results for author: Kaitlin Maile

Found 4 papers, 1 papers with code

Composable Function-preserving Expansions for Transformer Architectures

no code implementations11 Aug 2023 Andrea Gesmundo, Kaitlin Maile

Training state-of-the-art neural networks requires a high cost in terms of compute and time.

Equivariance-aware Architectural Optimization of Neural Networks

no code implementations11 Oct 2022 Kaitlin Maile, Dennis G. Wilson, Patrick Forré

Incorporating equivariance to symmetry groups as a constraint during neural network training can improve performance and generalization for tasks exhibiting those symmetries, but such symmetries are often not perfectly nor explicitly present.

Neural Architecture Search

When, where, and how to add new neurons to ANNs

1 code implementation17 Feb 2022 Kaitlin Maile, Emmanuel Rachelson, Hervé Luga, Dennis G. Wilson

Neurogenesis in ANNs is an understudied and difficult problem, even compared to other forms of structural learning like pruning.

Cannot find the paper you are looking for? You can Submit a new open access paper.