AI Paper Atlas

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

Telescoping Density-Ratio Estimation

Benjamin Rhodes, Kai Xu, Michael U. Gutmann

NeurIPS 2020 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Likelihood-free and graphical model estimation · 5 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
Replace a single density-ratio estimate with telescoping bridgesDensity-ratio estimation · 1d peaked ratioratio estimation error: 100 samples; 100,000 samplesImproved
Replace a single density-ratio estimate with telescoping bridgesDensity-ratio estimation · High-dimensional Gaussian MIMI estimation accuracy: 20 nats; 80 nats; 320 dimensional variablesImproved
Replace single-ratio baselines with TRESpatialMultiOmniglot representation learning · SpatialMultiOmniglotlinear evaluation accuracy: 20-60%; ~3%; >97%Improved
Replace NCE single-ratio estimation with TRE using Gaussian noiseMNIST EBM · MNISTbpd (RAISE): 1.99 -> 1.35Improved
Replace NCE single-ratio estimation with TRE using Gaussian Copula noiseMNIST EBM · MNISTbpd (RAISE): 1.48 -> 1.23Improved
Replace NCE single-ratio estimation with TRE using RQ-NSF noiseMNIST EBM · MNISTbpd (RAISE): 1.10 -> 1.09Improved
Generate linear-combination waymarks in x-space instead of z-space (ablation)TRE with Gaussian Copula noise · MNISTbpd (RAISE): 1.23 -> 1.33Hurt
Vary the number of waymarks between 5-30 (ablation)TRE with Gaussian Copula noise · MNISTapproximate log-likelihood (bpd): no greater than 0.03 bpdNo clear effect
Add an attention layer (ablation)TRE MNIST architecture · MNISTestimated log-likelihoodNo clear effect
Add spectral normalisation (ablation)TRE MNIST architecture · MNISTperformanceNo clear effect
Add Instance Normalisation (ablation)TRE energy-based modelling · Energy-based modellingperformanceHurt
Use average pooling rather than max pooling (ablation)TRE SpatialMultiOmniglot architecture · SpatialMultiOmniglotfinal performanceMixed

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