AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Non-Adversarial Mapping with VAEs

Yedid Hoshen

NeurIPS 2018 · Adversarial robustness and security › Attacks on text and graphs · approach: Defenses against universal perturbations · 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
replace per-example latent codes (NAM) with a learned stochastic VAE encoderNAM · MNIST->SVHNtranslated digit accuracy (%): 31.9 -> 51.7Improved
replace deterministic encoder (AE-NAM) with stochastic VAE encoder (ablation)VAE-NAM · MNIST->SVHNtranslated digit accuracy (%): 16.7 -> 51.7Improved
replace deterministic encoder (AE-NAM) with stochastic VAE encoder (ablation)VAE-NAM · SVHN->MNISTtranslated digit accuracy (%): 37.4 -> 37.4No clear effect
replace per-example latent codes (NAM) with VAE encoderNAM · SVHN->MNISTtranslated digit accuracy (%): 33.3 -> 37.4Improved
replace per-example latent codes (NAM) with VAE encoderNAM · Car2Carroot median residual error: 1.47 -> 1.38Improved
replace optimization-based inference (NAM) with feedforward encoderNAM · Evaluating 100 analogies, 64x64evaluation time: 23s -> 0.013sImproved
use stochastic VAE encoder instead of deterministic AE-NAM (ablation)VAE-NAM · edges2shoesrealism and sample variety (qualitative)Mixed

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