AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Processing of missing data by neural networks

Marek Śmieja, Łukasz Struski, Jacek Tabor, Bartosz Zieliński, Przemysław Spurek

NeurIPS 2018 · Architectures and efficiency › RNNs and normalization · approach: Representation geometry and collapse · 74 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
Represent missing inputs with a learned density and use expected first-layer neuron activations instead of single-value imputationAutoencoder on MNIST incomplete images · MNIST incomplete imagestotal reconstruction MSE: 0.01189 -> 0.01056Improved
Represent missing inputs with expected activations rather than k-nn imputationAutoencoder on MNIST incomplete images · MNIST incomplete imagesMSE inside mask: 0.00722 -> 0.00810Hurt
Represent missing inputs with expected activations rather than k-nn imputationAutoencoder on MNIST incomplete images · MNIST incomplete imagesMSE outside mask: 0.00468 -> 0.00246Improved
Use expected activations instead of MICE imputationMLP on ESR · ESR, 25% missingaccuracy: 0.823 -> 0.815Hurt
Use expected activations instead of MICE imputationMLP on ESR · ESR, 50% missingaccuracy: 0.816 -> 0.817Improved
Use expected activations instead of MICE imputationMLP on ESR · ESR, 75% missingaccuracy: 0.786 -> 0.787No clear effect
Use expected activations instead of MICE imputationMLP on ESR · ESR, 90% missingaccuracy: 0.670 -> 0.760Improved
Use expected activations rather than the strongest reported baselineRBFN on UCI datasets with internal missing attributes · hepatitisaccuracy: 0.825 -> 0.846Improved
Use expected activations rather than the strongest reported baselineRBFN on UCI datasets with internal missing attributes · kidney diseaseaccuracy: 0.995 -> 0.993Hurt
Use expected activations rather than the strongest reported baselineRBFN on UCI datasets with internal missing attributes · horseaccuracy: 0.826 -> 0.864Improved
Use expected activations rather than the strongest reported baselineRBFN on UCI datasets with internal missing attributes · pimaaccuracy: 0.769 -> 0.747Hurt
Use expected activations rather than the strongest reported baselineRBFN on UCI datasets with internal missing attributes · winconsinaccuracy: 0.970 -> 0.970No 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