Sitemap

A list of all the posts and pages found on the site. For you robots out there, there is an XML version available for digesting as well.

Pages

Posts

Faster Siamese Network Training and Inference

In this post, I look at a couple of ways you can dramatically speed up training of Siamese neural networks. These are two relatively simple tricks that I’ve not seen anywhere else which, when combined, can give a ~2.5x speed-up when training Siamese networks on GPU in PyTorch.

Visualising the Legendre Transform

The Legendre transform is incredibly useful in a wide range of areas including physics, economics, statistics and machine learning. However, its definition can be somewhat opaque. In this post, we look at how we might gain intuition for how the definition works by constructing it from scratch in a visual way.

portfolio

publications

Information Component Analysis for Facial Images

Wood, Danny (2014). "Information Component Analysis for Facial Images" MSc Thesis.

Effects of Network Weight Structure in Echo State Networks

Wood, Danny (2020). "Effects of Network Weight Structure in Echo State Networks." PhD Thesis.

My thesis explored memory and stability in Echo State Networks. Drawing on tools from control theory, I gave exact expressions for calculating the memory capacity of these networks. I also looked at the “different timescales” phenomenon in recurrent neural networks, characterising the phenomenon both in terms of the memory capacity of deep network layers and in terms of how the sensitivity of different layers to perturbations in input varies over time.

Two-stage Classification for Detecting Murmurs from Phonocardiograms Using Deep and Expert Features

Summerton, Sara, et al. "Two-stage Classification for Detecting Murmurs from Phonocardiograms Using Deep and Expert Features." Computing in Cardiology 2022: 49th Computing in Cardiology Conference. 2022.

Submitted as an entry in the 2022 Physionet Challenge as member of the team Murmur Mia! (winner of best team name). Hidden Markov Model segmentation code used in the paper can be found here.

Bias-Variance Decompositions for Margin Losses

Wood, Danny, Tingting Mu, and Gavin Brown. "Bias-Variance Decompositions for Margin Losses." International Conference on Artificial Intelligence and Statistics. PMLR, 2022.

A Unified Theory of Diversity in Ensemble Learning

Wood, Danny, et al. "A Unified Theory of Diversity in Ensemble Learning." arXiv preprint arXiv:2301.03962 (2023).

Model-agnostic variable importance for predictive uncertainty: an entropy-based approach

Wood, Danny, et al. "Model-agnostic variable importance for predictive uncertainty: an entropy-based approach" Data Mining and Knowledge Discovery (2024).

talks

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.