AI Paper Atlas

Research map / Learning theory and algorithms / Sample complexity

BR-SNIS: Bias Reduced Self-Normalized Importance Sampling

Gabriel Cardoso, Sergey Samsonov, Achille Thin, Eric Moulines, Jimmy Olsson

NeurIPS 2022 · Learning theory and algorithms › Sample complexity · approach: Optimal mean estimation and subsampling · 0 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
recycle i-SIR samples (BR-SNIS) instead of standard SNISSNIS · Gaussian mixture, M=16384, N in {129,513}bias: by a factor of almost 10Improved
recycle i-SIR samples (BR-SNIS) instead of standard SNISSNIS · Gaussian mixture, M=16384MSE: around 20%Hurt
use burn-in k0 = floor(0.625 kmax) instead of kmax-1 (ablation)BR-SNIS · Gaussian mixturebias and MSE: 3 times less bias, 10% more MSEMixed
BR-SNIS instead of SNIS, M=32SNIS · Bayesian logistic regression, CoverTypeTV distance: 0.0028 -> 0.0014Improved
BR-SNIS instead of SNIS, M=512SNIS · Bayesian logistic regression, CoverTypeTV distance: 0.0026 -> 0.0013Improved
BR-SNIS instead of SNIS, M=32SNIS · Bayesian logistic regression, BreastTV distance: 0.00011 -> 7.9e-5Improved
BR-SNIS instead of SNIS, M=512SNIS · Bayesian logistic regression, HeartTV distance: 7.8e-5 -> 4.9e-5Improved
BR-SNIS instead of SNISSNIS · Bayesian logistic regression, three datasetsposterior mean bias vs varianceMixed
train IWAE decoder with BR-SNIS gradient (k=4) instead of plain IWAE, d=20IWAE · binarized MNISTmean log-likelihood: -91.5 -> -90.5Improved
train IWAE decoder with BR-SNIS gradient (k=4) instead of plain IWAE, d=50IWAE · binarized MNISTmean log-likelihood: -92.9 -> -90.1Improved
increase bootstrap rounds k from 4 to 8, d=20 (ablation)BR-IWAE · binarized MNISTmean log-likelihood: -90.5 -> -90.4No clear effect
increase bootstrap rounds k from 4 to 8, d=50 (ablation)BR-IWAE · binarized MNISTmean log-likelihood: -90.1 -> -90.1No clear effect

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