AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Scaling Symbolic Methods using Gradients for Neural Model Explanation

Subham Sekhar Sahoo, Subhashini Venugopalan, Li Li, Rishabh Singh, Patrick Riley

ICLR 2021 · Architectures and efficiency › RNNs and normalization · approach: Theory of depth and width · 1 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 SMT minimization to sparsify the IG-selected input mask (ablation)SMUG · ImageNet validationSparsity%: 43.3 ± 0.32 -> 17.7 ± 0.10Improved
Use SMT minimization rather than the unminimized top-neuron receptive-field mask (ablation)SMUG · ImageNet validationLSC: -1.23 -> -1.26Improved
Use SMUG instead of pixel-space Integrated GradientsInception-v1 explanations · ImageNet validationLSC: -0.29 -> -1.26Improved
Increase the number of selected top-k neurons (ablation)SMUG · ImageNetmask sizeHurt
Increase the number of selected top-k neurons (ablation)SMUG · ImageNetsolver runtimeHurt
Decrease gamma (ablation)SMUG · ImageNetmask sizeImproved
Decrease gamma (ablation)SMUG · ImageNetsolver runtimeHurt
Encode the full network in SMT rather than selecting first-layer neurons with gradient informationSMT-based explanation on MNIST · MNIST validationsolved instances: 34 / 100Hurt

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