From f1431a53ad36cd281110358bf558ca5ac165e7f4 Mon Sep 17 00:00:00 2001 From: homerjed Date: Mon, 7 Oct 2024 17:37:30 +0200 Subject: [PATCH] paper --- paper/paper.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/paper/paper.md b/paper/paper.md index 98865fc..f15e093 100644 --- a/paper/paper.md +++ b/paper/paper.md @@ -100,7 +100,7 @@ $$ This ODE can be solved with an initial-value problem that maps a prior sample from a multivariate Gaussian to the data distribution. This inherits the formalism of continuous normalising flows [@neuralodes; @ffjord] without the expensive ODE simulations used to train these flows - this allows for a likelihood estimate based on diffusion models [@sde_ml]. -![A diagram showing a log-likelihood calculation over a 2D space within which a dataset of samples drawn from a Gaussian mixture model with eight components \label{fig:8gauss}](8gauss.png){ width=80% } +![A diagram showing a log-likelihood calculation over a 2D space within which a dataset of samples drawn from a Gaussian mixture model with eight components. The log-likelihood is calculated using the ODE and a trained diffusion model. \label{fig:8gauss}](8gauss.png){ width=50% } The likelihood estimate under a score-based diffusion model is estimated by solving the change-of-variables equation for continuous normalising flows.