AI Paper Atlas

Research map / Optimization / Optimizers

Stochastic Bouncy Particle Sampler

Ari Pakman, Dar Gilboa, David Carlson, Liam Paninski

ICML 2017 · Optimization › Optimizers · approach: Stochastic gradient MCMC · 24 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 SBPS with an adaptive local-regression thinning bound instead of optimally tuned stochastic-gradient samplersBayesian logistic regression · Bayesian logistic regressionmixing speedImproved
Use SBPS instead of exact-bound subsampling samplersBayesian logistic regression · Bayesian logistic regressionmixing speedImproved
Add diagonal adaptive preconditioningSBPS · Bayesian logistic regressionconvergence to posterior modeImproved
Estimate expectations along continuous trajectory segments rather than using discrete samplesSBPS · Continuous trajectory sampling, when r/b is less than 1expectation-estimation errorImproved
Use SBPS rather than SGLD with a selected step sizeMNIST posterior sampling · MNISTmixingNo clear effect
Use SBPS rather than large-step SGLD or mSGNHTHighly non-Gaussian hyperboloid posterior · Hyperboloid posteriorposterior explorationImproved
Use zero-mean independent noisy gradientsBPS · BPS with noisy gradientsinvariant-distribution biasNo clear effect
Use noisy rather than exact gradientsBPS · BPS with noisy gradientsmixing speed per iterationHurt
Replace the local linear predictive model with Gaussian process regression (ablation)SBPS · Examples consideredmixing speedNo clear effect
Use a small, controlled bias through approximate thinning rather than exact conservative boundsSBPS · Bayesian logistic regressionmixing speedImproved
Estimate expectations by continuous trajectory integration rather than discrete samplingSBPS · Continuous trajectory sampling, when r/b is approximately 1expectation-estimation errorNo clear effect
Use a poor SGLD step size rather than an appropriately chosen oneSGLD · MNIST posterior samplingmixingMixed

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