GNN XAI 学习⼤纲
Papers
XAI papers
“Why Should I Trust You?”: Explaining the Predictions of Any Classifier, KDD 2016, Marco Tulio Ribeiro et
al.
Note：本⽂提出的LIME是最经典的XAI post analysis⽅法。
A Unified Approach to Interpreting Model Predictions, NIPS 2017, Scott M. Lundberg et al.
Note: 该论⽂的⽹⻚中还列出和对⽐了其他常⻅的多种XAI⽅法，相关论⽂也请仔细阅读并⽐较。
Beyond Accuracy: Behavioral Testing of NLP models with CheckList, ACL 2020, Marco Tulio Ribeiro et al.
Note: LIME作者另⼀篇XAI论⽂，获得ACL2020最佳论⽂。
A Survey of the State of Explainable AI for Natural Language Processing, AACL 2020
Towards better understanding of gradient-based attribution methods for Deep Neural Networks, ICLR
2018
Learning to Explain: An Information-Theoretic Perspective on Model Interpretation, ICML 2018
L-Shapley and C-Shapley: Efficient Model Interpretation for Structured Data, ICLR 2019
还有⼏篇NeurIPS2020收录的以及 ICLR2021 submission的论⽂稍后列出。
Rationale-based XAI papers
Rationalizing Neural Predictions, EMNLP 2016, Tao Lei et al.
Note：本⽂为基于Rationale⽅法的开⼭之作。
Interpretable Neural Predictions with Differentiable Binary Variables, ACL 2019, Jasmijn Bastings, Wilker
Aziz, Ivan Titov
Rationalizing Text Matching: Learning Sparse Alignments via Optimal Transport, ACL 2020, Tao Lei et al.
Note：本⽂是Tao Lei的另外⼀篇最新相关论⽂，所以也列出参阅。
How do Decisions Emerge across Layers in Neural Models? Interpretation with Differentiable Masking,
EMNLP 2020, Nicola De Cao, Michael Sejr Schlichtkrull, Wilker Aziz, Ivan Titov
Learning to Faithfully Rationalize by Construction, ACL 2020, Sarthak Jain et al.
Note：作者声称是对Tao Lei EMNLP2016论⽂的改进。
Towards Explainable NLP: A Generative Explanation Framework for Text Classification, Hui Liu, Qingyu
Yin, William Yang Wang
Towards Interpretable Natural Language Understanding with Explanations as Latent Variables, NeurIPS
2020
Interpreting Image Classifiers by Generating Discrete Masks, TPAMI 2020, Hao Yuan et al.
Note: 将基于Rationale的⽅法⽤于图像，输出hard mask。
GNN XAI Papers
GNNExplainer: Generating Explanations for Graph Neural Networks, NeurIPS 2019, Rex Ying, Dylan
Bourgeois, Jiaxuan You, Marinka Zitnik, Jure Leskovec
Note: Stanford发的第⼀篇GNN XAI的论⽂。
XGNN: Towards Model-Level Explanations of Graph Neural Networks, KDD 2020, Hao Yuan et al.
Note：TAMU通过⽣成实现最⼤化预测值的input graph来解释GNN模型。
Multi-Objective Molecule Generation using Interpretable Substructures, ICML 2020, Wengong Jin, Regina
Barzilay, Tommi Jaakkola
Note: MIT新药发现研究
Drug discovery with explainable artificial intelligence, Nature Machine Intelligence volume 2, pages573–
584(2020)
A Deep Learning Approach to Antibiotic Discovery, Cell 2020
Parameterized Explainer for Graph Neural Network, NeurIPS 2020, Dongsheng Luo, Wei Cheng, Dongkuan
Xu, Wenchao Yu, Bo Zong, Haifeng Chen, Xiang Zhang
Hard Masking for Explaining Graph Neural Networks, ICLR 2021 Submission
Interpreting Graph Neural Networks for NLP With Differentiable Edge Masking, ICLR 2021 Submission,
Michael Sejr Schlichtkrull, Nicola De Cao, Ivan Titov
Recent tutorials
XAI tutorials
Tutorial on Explaining ML Predictions: State-of-the-art, Challenges, and Opportunities, NeurIPS 2020,
Himabindu Lakkaraju, Julius Adebayo, Sameer Singh
Tutorial on Interpreting Predictions of NLP Models, EMNLP 2020, Eric Wallace, Matt Gardner, Sameer
Singh
Interpretability and Analysis in Neural NLP, ACL 2020
Interpreting and Explaining Deep Neural Networks: A Perspective on Time Series Data, KDD 2020
Human-Centered Explainability for Healthcare, KDD 2020
GNN tutorials
Advanced Deep Graph Learning: Deeper, Faster, Robuster, and Unsupervised, KDD 2020
Multi-modal Network Representation Learning: Methods and Applications, KDD 2020
Recent Advances on Graph Analytics and Its Applications in Healthcare, KDD 2020
Graph Neural Networks: Models and Applications, AAAI 2020/2021
GNN for NLP tutorials
Graph Neural Networks in NLP, CCL2019, Yue Zhang
A Tutorial on Graph Neural Networks for Natural Language Processing, EMNLP2019
Note: 上⾯⼤部分GNN for NLP的⽅法在后BERT时代都显得过时了。
Resources
Courses on XAI
Interpretability and Explainability in Machine Learning, COMPSCI 282BR, Harvard University by Hima
Lakkaraju
Courses on GNN
CS224W: Machine Learning with Graphs, Jure Leskovec
Talks on XAI
DARPA’s Explainable Artificial Intelligence (XAI) Program, Dave Gunning
Please Stop Doing “Explainable” ML, Cynthia Rudin
Talks on GNN
AI Cures Drug Discovery Conference, MIT
Talk by Lee Hungyi
Generation by RL and GAN, and video
GNN books
Deep Learning on Graphs, Yao Ma and Jiliang Tang
Graph Representation Learning Book, William L. Hamilton, McGill University
Written with StackEdit.
