Robust Interpretation

Robust interpretation in machine learning aims to create explanations for model predictions that are reliable and consistent, even when faced with slight input variations or adversarial attacks. Current research focuses on improving the faithfulness and stability of interpretation methods for various architectures, including vision transformers and convolutional neural networks, often employing techniques like adversarial training, causal analysis, and diffusion-based denoising to enhance robustness. This pursuit is crucial for building trust in AI systems across diverse applications, particularly in high-stakes domains like medical diagnosis and autonomous driving, where understanding model decisions is paramount for safety and accountability.

Papers

October 10, 2024

Unlearning-based Neural Interpretations
Ching Lam Choi, Alexandre Duplessis, Serge Belongie
Natural Gradient Decision Boundary Salient Feature Efficient Baseline Robust Interpretation

April 30, 2024

Enhancing Deep Learning Model Explainability in Brain Tumor Datasets using Post-Heuristic Approaches
Konstantinos Pasvantis, Eftychios Protopapadakis
Deep Learning Model Line by Line Explanation High Explainability Brain Tumor Robust Interpretation LIME Explanation

November 29, 2023

Improving Interpretation Faithfulness for Vision Transformers
Lijie Hu, Yixin Liu, Ninghao Liu, Mengdi Huai, Lichao Sun, Di Wang
Vision Transformer Neural Architecture Denoising Diffusion Self Attention Network Robust Interpretation Stable Explanation

July 4, 2023

Interpretable Computer Vision Models through Adversarial Training: Unveiling the Robustness-Interpretability Connection
Delyan Boychev
Adversarial Attack Native Robustness Adversarial Training Inherent Interpretability Robust Model Robust Interpretation Interpretable Computer Vision

May 15, 2023

Causal Analysis for Robust Interpretability of Neural Networks
Ola Ahmad, Nicolas Bereux, Loïc Baret, Vahid Hashemi, Freddy Lecue
Neural Network Causal Graph Causal Analysis Robust Interpretation

January 26, 2023

Certified Interpretability Robustness for Class Activation Mapping
Alex Gu, Tsui-Wei Weng, Pin-Yu Chen, Sijia Liu, Luca Daniel
Native Robustness Class Activation Robustness Guarantee Robust Interpretation

January 8, 2023

MoreauGrad: Sparse and Robust Interpretation of Neural Networks via Moreau Envelope
Jingwei Zhang, Farzan Farnia
Neural Network Many Sparse Robust Interpretation Moreau Envelope Gradient Based Interpretation

November 29, 2022

Towards More Robust Interpretation via Local Gradient Alignment
Sunghwan Joo, Seokhyeon Jeong, Juyeon Heo, Adrian Weller, Taesup Moon
Feature Attribution Attribution Method Feature Attribution Method Gradient Alignment Robust Interpretation

August 19, 2022

SAFARI: Versatile and Efficient Evaluations for Robustness of Interpretability
Wei Huang, Xingyu Zhao, Gaojie Jin, Xiaowei Huang
Native Robustness Inherent Interpretability Explainable AI Versatile Approach Efficient Evaluation Robust Interpretation Gradient Based XAI

February 21, 2022

Ligandformer: A Graph Neural Network for Predicting Compound Property with Robust Interpretation
Jinjiang Guo, Qi Liu, Han Guo, Xi Lu
Graph Neural Network Quantitative Structure Activity Relationship Chemical Structure Robust Interpretation Compound Property

November 9, 2021

Self-Interpretable Model with TransformationEquivariant Interpretation
Yipei Wang, Xiaoqian Wang
Robust Interpretation Self Interpretable Model