atlas

Research map / Detection and segmentation / CNN features

Sanity Checks for Saliency Maps

NeurIPS 2018 · Detection and segmentation › CNN features · approach: RGB salient object detection · 580 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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
cascading randomization of model weightsGradient saliency · Inception v3 ImageNet, CNN/MLP MNIST and Fashion MNISTSpearman rank correlation with original mapHurt
cascading randomization of model weightsGuided BackProp and Guided GradCAM · Inception v3 ImageNet, MNIST, Fashion MNISTsimilarity to original mapNo clear effect
cascading randomization of model weightsGradCAM · Inception v3 ImageNetsimilarity to original mapHurt
cascading randomization of model weightsIntegrated Gradients and Gradient x Input · Inception v3 ImageNetSpearman rank correlation (abs vs no abs)Mixed
train on randomly permuted labelsGradient and SmoothGrad · MNIST, Fashion MNISTrank correlation with true-label model mapHurt
train on randomly permuted labelsGuided BackProp · MNISTvisual plausibility of mapNo clear effect
independent single-layer randomizationGuided BackProp and Guided GradCAM · Inception v3, CNN, MLPsimilarity to original mapNo 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