AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Active Learning of Conditional Generative Models via the Transport Neural Tangent Kernel

Jayoung Ryu, Kyunghyun Cho, Romain Lopez

NeurIPS 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Score-based generative modeling analysis · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes an active learning method for conditional generative models of distributional experimental outputs, using a transport neural tangent kernel to guide experiment selection.

Problem

Automated experimental design for high-throughput experiments with distributional outputs, where acquiring data at new conditions is expensive and existing active learning frameworks make restrictive assumptions about the output distribution.

Approach

Introduces a transport neural tangent kernel to active learning of conditional generative models, enabling selection of new experimental conditions for distributional outputs rather than assuming simpler output models.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use tNTK instead of mean-based NTK for acquisitionGP total-variance active learning · Gaussian mixture (quantile, 5+1)change in Wasserstein test risk (x1e-3): -416.43 -> -429.52Improved
use tNTK instead of mean-based NTK for acquisitionGP total-variance active learning · Dew Point (quantile, 5+1)change in Wasserstein test risk (x1e-3): -178.44 -> -221.18Improved
use tNTK instead of NTKconditional flow matching AL · Radial flow, CLINC150, Van der Polchange in Sinkhorn divergence: CLINC150 5+1: -306.01 -> -672.85Improved
use tNTK instead of NTK at a larger budgetconditional flow matching AL · CLINC150 (5+5)change in Sinkhorn divergence: -884.02 vs -852.31No clear effect
use tNTK instead of NTKconditional Monge gap AL · K562-Chemical and BMDM-Geneticchange in Sinkhorn divergence: BMDM 20+10: -130.32 -> -144.90Improved
use tNTK instead of NTK at small budgetconditional Monge gap AL · K562-Chemical (10+1)change in Sinkhorn divergence: -9.33 -> -8.65No clear effect
use tNTK instead of mean-based NTKaffine Gaussian transport AL · Synthetic Gaussian 2DWasserstein loss: 1.2x lower than randomImproved
use Wasserstein-risk acquisition instead of mean-only acquisition (ablation)1D GP active learning · Synthetic 1D conditional densityWasserstein riskImproved
use NTK on mean AL instead of random1D GP active learning · Synthetic 1D conditional densityshape errorNo clear effect
use NTK from mean-var or quartile networks instead of mean onlyNTK baseline · Quantile datasetschange in Wasserstein test risk: Gaussian mixture 5+1: -416.43 vs -367.99No clear effect
use BALSA-EMD instead of randomactive learning baseline · Dew Point (quantile, 5+1)change in Wasserstein test risk (x1e-3): -79.90 vs -79.51No clear effect
use tNTK instead of randomconditional Monge gap AL · K562-Chemical (10+1)change in Sinkhorn divergence: -9.56 -> -8.65Hurt

Experimental setup

Builds on: neural tangent kernel

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy