AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Efficiently Computing Compact Formal Explanations

Min Wu, Xiaofu Li, Haoze Wu, Clark Barrett

AAAI 2026 · Learning theory and algorithms › Prediction and decision losses · approach: Provable post-hoc explanations · 13 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 saliency order with bound propagation order (ablation)VeriX+ binary search · GTSRB-CNNexplanation size: 569.0 -> 355.8Improved
use binary search traversal instead of sequential (ablation)VeriX · MNIST-CNNgeneration time: 398.9 -> 46.3Improved
add confidence ranking to CHECK (ablation)VeriX · MNIST-CNNgeneration time: 439.2 -> 398.9Improved
use QuickXplain instead of binary search (ablation)VeriX+ · GTSRB-CNNexplanation size: 355.8 -> 355.3 size, 251.0 -> 330.1 sMixed
use bound propagation order instead of saliency (ablation)VeriX+ binary search · MNIST-CNNexplanation size: 105.7 -> 100.8Improved
use bound propagation order instead of saliency (ablation)VeriX+ binary search · CIFAR10-CNNexplanation size: 664.8 -> 553.5Improved
adversarial training of MNIST-FCMNIST-FC · MNIST-FC original correct samplesexplanation size: 177.20 -> 128.22Improved
adversarial training of MNIST-FCMNIST-FC · MNIST-FC original correct samplesepsilon-robust rate: 4% -> 52.3%Improved
use explanation size instead of max softmax for OOD detectionMNIST-CNN · CIFAR10 OODAUROC: 93%Improved
use VeriX+ QuickXplain instead of Izza et al. Dichotomic searchVeriX+ · MNIST-FCgeneration time: 82.3 -> 35.6Improved

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