← Back to the feed
AI & softwareResearch findings

TorchCurves: Learnable Parametric Curves as Differentiable Primitives

Original published 2026.11.11 03:58 KSTAdded to Blackdeck 2026.11.12 08:53 KST

What has changed?

Supports differentiation through both inputs and curve coefficients, enabling differentiable parametric curve learning.

Key points

  1. TorchCurves presents vector-valued parametric curves as learnable primitives in PyTorch.
  2. Differentiation through both inputs and coefficients enables applications like continuous feature embeddings.
  3. Curve coefficients can be optimized jointly with surrounding models or generated by other networks.

Connect it to your work

Treats vector-valued parametric curves as learnable primitives with B-spline and Legendre polynomial bases.

What remains unproven

This research note is based on an abstract; check the original for test conditions and application scope.

Originals used for this note

Links open the original publisher in a new tab.

  1. arxivTorchCurves: Differentiable Parametric Curves