# Neurips 2026 - Active Learning Paper

**URL:** <https://melissa.discourse.group/t/neurips-2026-active-learning-paper/71>\
**Category:** General\
**Created:** [November 13, 2026, 10:57am UTC](https://melissa.discourse.group/t/neurips-2026-active-learning-paper/71 "2026-11-13T10:57:36Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![bruno.raffin](https://avatars.discourse-cdn.com/v4/letter/b/94ad74/32.png) [@bruno.raffin](https://melissa.discourse.group/u/bruno.raffin)\
**Post date:** [November 13, 2026, 10:57am UTC](https://melissa.discourse.group/t/neurips-2026-active-learning-paper/71/1 "2026-11-13T10:57:36Z")

</div>

The Melisa team has an accepted paper at Neurips 2026: [Learning Where to Simulate: Generative Active Sampling for Online PDE Surrogate Training](https://melissa-sa.github.io/ogas/). The paper is about improving neural surogate training by adding a secondary diffusion network that learns to sample the parameters of the next solver instances to run such that this solver generates hard-to-learn samples. Full online workflow based on Melissa.