New Transport Neural Tangent Kernel Enhances Active Learning for Generative Models

Researchers have developed a new active learning framework for conditional generative models, addressing the limitations of existing methods that focus solely on scalar or vector-valued outputs. The proposed transport Neural Tangent Kernel (tNTK) captures both the mean and shape of predicted distributions, unlike traditional mean-based active learning. This kernel, which admits a closed-form expression for affine Gaussian transport and tractable Monte Carlo estimators, allows for a distribution-aware active learning strategy. Experiments on various conditional generative models and datasets, including intent-labeled sentences and single-cell perturbation responses, show that the tNTK outperforms baseline methods. AI

IMPACT Enhances active learning strategies for generative models, potentially improving data efficiency in scientific experiments.

RANK_REASON The cluster contains a research paper detailing a new kernel method for active learning in generative models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Active Learning
• affine Gaussian transport
• arXiv
• Conditional Flow Matching
• Conditional Generative Models
• conditional Monge gap
• Gaussian process
• Hugging Face
• Neural tangent kernel
• Transport Neural Tangent Kernel
• Wasserstein

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →