AI Paper Atlas

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

Understanding Instance-based Interpretability of Variational Auto-Encoders

Zhifeng Kong, Kamalika Chaudhuri

NeurIPS 2021 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Disentangled and discrete latent models · 9 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 VAE-TracIn self-influence ranking to flag extra samplesβ-VAE · MNIST with EMNIST extra samplesdetection AUC: 0.887Improved
use VAE-TracIn self-influence ranking to flag extra samplesβ-VAE · CIFAR-10 with CelebA extra samplesdetection AUC: 0.760Improved
increase latent dimension from 64 to 128 (ablation)VAE-TracIn · MNISTself-influence top-1 frequency: 0.992 -> 1.000Improved
increase latent dimension from 64 to 128 (ablation)VAE-TracIn · CIFAR-10self-influence top-1 frequency: 0.609 -> 0.602Hurt
increase latent dimension from 64 to 128 (ablation)VAE-TracIn · MNIST synthesized major/minor componentsmajor-component top-1 score: 0.422 -> 0.467Improved
replace perceptual autoencoder nearest-neighbor ranking with VAE-TracInMNIST synthesized major/minor components · MNIST synthesized major/minor componentsmajor-component top-1 score: 0.633 -> 0.422Hurt
replace perceptual autoencoder nearest-neighbor ranking with VAE-TracInMNIST synthesized major/minor components · MNIST synthesized major/minor componentsminor-component rank 25% quantile: 6943 -> 1097Improved
estimate influence with VAE-TracInβ-VAE · CIFAR-10 subclassesPearson correlation with loss change: 0.882, 0.910, 0.895, 0.934, 0.828, 0.865, 0.908, 0.852, 0.861, 0.613Improved
use VAE-TracIn to identify the strongest self-proponentβ-VAE · CIFAR-10 subclasses (average)self-influence top-1 frequency: 0.998Improved
use VAE-TracIn to identify the strongest self-proponentβ-VAE · CIFAR-10self-influence top-1 frequency: 0.609; 0.602Mixed

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