mlsanityData to 14 Oct 2026

Generalizing Stochastic Smoothing for Differentiation and Gradient Estimation

arXiv 2410.08125 ↗First posted 10 Oct 2024cs.LG stat.MLcited by 4

Not on the map. Why?

Abstract (arXiv)

We deal with the problem of gradient estimation for stochastic differentiable relaxations of algorithms, operators, simulators, and other non-differentiable functions. Stochastic smoothing conventionally perturbs the input of a non-differentiable function with a differentiable density distribution with full support, smoothing it and enabling gradient estimation. Our theory starts at first principles to derive stochastic smoothing with reduced assumptions, without requiring a differentiable density nor full support, and we present a general framework for relaxation and gradient estimation of non-differentiable black-box functions f:ℝn→ℝm. We develop variance reduction for gradient estimation from 3 orthogonal perspectives. Empirically, we benchmark 6 distributions and up to 24 variance reduction strategies for differentiable sorting and ranking, differentiable shortest-paths on graphs, differentiable rendering for pose estimation, as well as differentiable cryo-ET simulations.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 29 of its 57 references that are in the map. First posted on arXiv 10 Oct 2024.

Builds on

29 of 57 references.

References · 29

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  2. 2013Sinkhorn Distances: Lightspeed Computation of Optimal Transportation Distancescited by 2,316
  3. 2017Neural 3D Mesh Renderercited by 437
  4. 2019Soft Rasterizer: A Differentiable Renderer for Image-based 3D Reasoningcited by 420
  5. 2017Soft-DTW: a Differentiable Loss Function for Time-Seriescited by 271
  6. 2019Differentiation of Blackbox Combinatorial Solverscited by 226
  7. 2020Fast Differentiable Sorting and Rankingcited by 197
  8. 2019Differentiable Ranks and Sorting using Optimal Transportcited by 113
  9. 2019Stochastic Optimization of Sorting Networks via Continuous Relaxationscited by 111
  10. 2022Do Differentiable Simulators Give Better Policy Gradients?cited by 89
  11. 2020Learning with Differentiable Perturbed Optimizerscited by 81
  12. 2019A Theoretical and Empirical Comparison of Gradient Approximations in Derivative-Free Optimizationcited by 68
17 more
  1. 2020SoftSort: A Continuous Relaxation for the argsort Operatorcited by 66
  2. 2021Differentiable Sorting Networks for Scalable Sorting and Ranking Supervisioncited by 31
  3. 2022Automatic Differentiation of Programs with Discrete Randomnesscited by 29
  4. 2023Fast, Differentiable and Sparse Top-k: a Convex Analysis Perspectivecited by 26
  5. 2022Monotonic Differentiable Sorting Networkscited by 25
  6. 2021Learning with Algorithmic Supervision via Continuous Relaxationscited by 21
  7. 2019Pix2Vex: Image-to-Geometry Reconstruction using a Smooth Differentiable Renderercited by 18
  8. 2023Differentiable Clustering with Perturbed Spanning Forestscited by 14
  9. 2023Differentiable Sampling of Categorical Distributions Using the CatLog-Derivative Trickcited by 14
  10. 2023Learning by Sorting: Self-supervised Learning with Group Ordering Constraintscited by 13
  11. 2021Storchastic: A Framework for General Stochastic Automatic Differentiationcited by 11
  12. 2022Learning with Differentiable Algorithmscited by 10
  13. 2023GeoPhy: Differentiable Phylogenetic Inference via Geometric Gradients of Tree Topologiescited by 10
  14. 2024Uncertainty Quantification via Stable Distribution Propagationcited by 10
  15. 2021Differentiable Rendering with Perturbed Optimizerscited by 7
  16. 2023Branches of a Tree: Taking Derivatives of Programs with Discrete and Branching Randomness in High Energy Physicscited by 6
  17. 2021Style Agnostic 3D Reconstruction via Adversarial Style Transfercited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 1; 2025: 1; 2026: 2.

4 papers.

  1. 2024Newton Losses: Using Curvature Information for Learning with Differentiable Algorithmscited by 2
  2. 2025A Start To End Machine Learning Approach To Maximize Scientific Throughput From The LCLS-II-HEcited by 0
  3. 2026Dimensional Peeking for Low-Variance Gradients in Zeroth-Order Discrete Optimization via Simulationcited by 0
  4. 2026Learning to Approximate Uniform Facility Location via Graph Neural Networkscited by 0