Explainable Artificial Intelligence.

可解释的人工智能

Posted by Jing on May 6, 2021

Let you know the principle of AI models from excellent works

This post is the collection of explainable AI techniques, tools, applications and reviews.

✅ Techniques

Class Activation maps
Anchors
Prediction Difference Analysis
Contextual Prediction Difference Analysis
Shapley Value Sampling
DeConvNet
Gradient/Saliency Maps
Gradient * input
FullGrad
GradCAM
Guided GradCAM
SmoothGrad
VarGrad
Integrated Gradients
Expressive gradients
Gradient SHAP
Deep SHAP
SHAP Interaction Index
Kernel SHAP
DeepLIFT SHAP
DeepLIFT
Excitation Backprop
Guided backpropagation
NeuronGuidedBackprop
Tree Explainer
GNNExplainer
GNN-LRP
Layer-wise Relevance Propagation
Spectral Relevance Analysis
DeepTaylor
LayerConductance
Local Rule-based Explanations
LIME 📝Introduction
STREAM
RISE
PatternNet
Pattern Attribution
Occlusion
Meaningful Perturbation
ExtremalPerturbation
Internal Influence
Representation Erasure
FIDO
NeuronConductance
TotalConductance
DeepDreams
TCAV
TCAV with RCV
UBS
SENN

✅ Tools & Applications

🔸 Heatmapping[web]
🔸 CNN-explainer[web]
🔸 Explainable AI Demos[web]
🔸 A Neural Network Playground[web]
🔸 Summit[web]
🔸 NeuralDivergence[web]
🔸 SCIN[web]
🔸 Neuroscope[software]
🔸 LUCID[library]
🔸 Keras-vis[library]
🔸 DeepExplain[library]
🔸 iNNvestigate[library]
🔸 TensorFlow Graph Visualizer[library]
🔸 tf-explain[library]
🔸 TorchRay[library]
🔸 Captum[library]
🔸 Explainable AI[commercial]
🔸 Explainable AI Platform[commercial]
🔸 exAID[commercial]
🔸 DASL[commercial]

✅ Reviews

🔹 Interpretable Machine Learning A Guide for Making Black Box Models Explainable. Christoph Molnar
🔹 Towards Robust Interpretability with Self-Explaining Neural Networks, 2018
🔹 Towards Explainable Artificial Intelligence, 2019
🔹 On the Integration of Knowledge Graphs into Deep Learning Models for a More Comprehensible AI—Three Challenges for Future Research, 2020
🔹 Explainable Deep Learning: A Field Guide for the Uninitiated, 2020
🔹 A Survey on Explainable Artificial Intelligence (XAI): Towards Medical XAI, 2020
🔹 Explainable Artificial Intelligence (xAI) Approaches and Deep Meta-Learning Models, 2020
🔹 Explainable Deep Learning Models in Medical Image Analysis, 2020
🔹 Achievements and Challenges in Explaining Deep Learning based Computer-Aided Diagnosis Systems, 2020
🔹 Interpretability and Explainability: A Machine Learning Zoo Mini-tour, 2020
🔹 Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications, 2021
🔹 Awesome machine learning interpretability, up to date

✅ Evaluations

The (Un)reliability of Saliency Methods: input variant
Sanity Checks for Saliency Maps: model and data
Evaluating the Visualization of What a Deep Neural Network Has Learned: AOPC
Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead, Cynthia Rudin