mastodontech.de ist einer von vielen unabhängigen Mastodon-Servern, mit dem du dich im Fediverse beteiligen kannst.
Offen für alle (über 16) und bereitgestellt von Markus'Blog

Serverstatistik:

1,4 Tsd.
aktive Profile

#neurips2023

0 Beiträge0 Beteiligte0 Beiträge heute

#neurips2023 is good and all but the AV here is absolutely awful. No idea if it’s the New Orleans conference center or the conference organizers at fault, but it’s all blurred slides and echoes and loud buzzing hums (at least I truly hope it’s not just my fading vision and tinnitus)

I am attending #NeurIPS for the first time, and in New Orleans for the first time. Which of these two things (#neurips2023 or NO) is the wilder? Sheer force of numbers suggests that the city has to outweigh the conference in chaos potential. And yet… this conference is really amazing. Genuinely stumped on this question (but will be mostly attending the conference and not hanging out in the French Quarter)

We propose a new family of probability densities that have closed form normalising constants. Our densities use two layer neural networks as parameters, and strictly generalise exponential families. We show that the squared norm can be integrated in closed form, resulting in the normalizing constant. We call the densities Squared Neural Family (#SNEFY), which are closed under conditioning.

Accepted at #NeurIPS2023. #MachineLearning #Bayesian #GaussianProcess

arxiv.org/abs/2305.13552

arXiv.orgSquared Neural Families: A New Class of Tractable Density ModelsFlexible models for probability distributions are an essential ingredient in many machine learning tasks. We develop and investigate a new class of probability distributions, which we call a Squared Neural Family (SNEFY), formed by squaring the 2-norm of a neural network and normalising it with respect to a base measure. Following the reasoning similar to the well established connections between infinitely wide neural networks and Gaussian processes, we show that SNEFYs admit a closed form normalising constants in many cases of interest, thereby resulting in flexible yet fully tractable density models. SNEFYs strictly generalise classical exponential families, are closed under conditioning, and have tractable marginal distributions. Their utility is illustrated on a variety of density estimation and conditional density estimation tasks. Software available at https://github.com/RussellTsuchida/snefy.