Below are some notes and papers I have written about deep learning.
notes
- Backpropping through Cholesky 23 May 2024
papers (google scholar)
- Approximate Muon with low-rank adapters
- Controlling changes to attention logits
- Scale-invariant attention
- Function-space learning rates
- Stochastic kernel regularisation improves generalisation in deep kernel machines
- Flexible infinite-width graph convolutional neural networks
- Convolutional deep kernel machines
- An improved variational approximate posterior for the deep Wishart process
- A theory of representation learning gives a deep generalisation of kernel methods