AI Paper Atlas

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

Automatic Differentiation of Programs with Discrete Randomness

Gaurav Arya, Moritz Schauer, Frank Schäfer, Christopher Vincent Rackauckas

NeurIPS 2022 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Gradient estimation for discrete variables · 11 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use stochastic derivatives instead of score function estimator for binomial variablegradient estimation · Bin(n,p)estimator variance order: n^3 -> nImproved
use stochastic triples instead of score function estimator (with or without control variate)inhomogeneous random walk · random walk, f(x)=x^2gradient estimator varianceImproved
differentiate particle filter resampling step with smoothed stochastic derivatives instead of ignoring itparticle filter likelihood gradient · hidden Markov model, d=2bias vs Kalman filter derivativeImproved
use smoothed instead of unsmoothed stochastic derivatives for non-linear programsstochastic derivative AD · Geo(p)^3 at p=0.01bias: <0.5% biasNo clear effect
switch particle filter AD from forward to reverse modeparticle filter · hidden Markov modelrun timeMixed
compare smoothed stochastic derivative variance to biased estimators in particle filterparticle filter · hidden Markov modelvarianceHurt

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