Noisy Feature Mixup ([preprint](https://arxiv.org/pdf/2110.02180.pdf)) and Long Expressive Memory for Sequence Modeling ([preprint](https://arxiv.org/pdf/2110.04744.pdf)).
Noisy Recurrent Neural Networks ([preprint](https://arxiv.org/pdf/2102.04877.pdf)), and Compressing Deep ODE-Nets using Basis Function Expansions ([preprint](https://arxiv.org/pdf/2106.10820.pdf)) which is joint work with Google Research.
Lipschitz Recurrent Neural Networks ([preprint](https://openreview.net/pdf?id=-N7PBXqOUJZ)) and Adversarially-Trained Deep Nets Transfer Better ([preprint](https://openreview.net/pdf?id=ijJZbomCJIm)).
Forecasting sequential data using consistent Koopman autoencoders ([preprint](https://arxiv.org/pdf/2003.02236.pdf)) and Error Estimation for Sketched SVD via the Bootstrap ([preprint](https://arxiv.org/pdf/2003.04937.pdf)).