Techniques for interpretable machine learning

By Mengnan Du, Ninghao Liu, Xia Hu. 2018

1,244 citations recorded by the source corpus.

Uncovering the mysterious ways machine learning models make decisions.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Causal Interpretability for Machine Learning - Problems, Methods and Evaluation (2020, background): “[18] provide a comprehensive survey of existing interpretable methods and discuss issues that should be considered in future work.”
A Survey of Fake News (2020, methodology): “Facilitating interpretability has been of great interest in artificial intelligence [Vilone and Longo 2020] and ML [Du et al. 2018] research.”
Are Explanations Helpful? A Comparative Study of the Effects of Explanations in AI-Assisted Decision-Making (2021, methodology): “havior of the entire AI model is categorized as global explanations, while methods that provide reasons for specic model predictions are categorized as local explanations [2, 24, 26].”
A comprehensive taxonomy for explainable artificial intelligence: a systematic survey of surveys on methods and concepts (2021, background): “2019 [41] B Perturbation based attention visualization 3”
Machine Learning Interpretability: A Survey on Methods and Metrics (2019, background): “The most significant one is the opaqueness or the lack of transparency [6], which inherently characterizes black box ML models.”
Bridging the Gap Between Ethics and Practice (2020, background): “In addition to the distinctions between intrinsically understandable models, post hoc , and prospective explanations, Du et al. [28] follow other researchers in distinguishing between global explanations that give an overview of what the algorithm does and local explanations that deal with specific…”
SurvNAM: The machine learning survival model explanation (2021, methodology): “These methods assume that contribution of a feature can be determined by measuring how prediction score changes when the feature is altered [45].”
Post-hoc Interpretability for Neural NLP: A Survey (2021, background): “Because of this relation, it is also often referred to as white-box models (Rudin, 2019; Du et al., 2019; Chatzimparmpas et al., 2020).”
Natural language processing applied to mental illness detection: a narrative review (2022, background): “The evaluation of a successful model does not only rely on performance, but also on its interpretability 197 , which is signi ﬁ cant for guiding clinicians to understand not only what has been extracted from text but the reasoning underlying some prediction 198 – 200 .”
Interpretable Machine Learning - A Brief History, State-of-the-Art and Challenges (2020, neutral): “rve as stand-alone ML algorithms, but also as building blocks for many IML approaches. 3 Today IML has reached a rst state of readiness. Research-wise, the eld is maturing in terms of methods surveys [75,41,120,96,1,6,23,15], further consolidation of terms and knowledge [42,22,82,97,88,17], and work”
DTCA: Decision Tree-based Co-Attention Networks for Explainable Claim Verification (2020, background): “Explainable Machine Learning Our work is also related to explainable machine learning, which can be generally divided into two categories: intrinsic explainability and post-hoc explainability (Du et al., 2018).”
Towards Explanation of DNN-based Prediction with Guided Feature Inversion (2018, background): “Introduction DNN interpretation techniques can be grouped into three categories [1]: •Design interpretable network architectures •Post-hoc interpret a pre-trained model •Post-hoc explain a prediction of a pre-trained model”
Explainability in Graph Neural Networks: A Taxonomic Survey (2020, background): “Recent survey works [23], [24], [25], [26] provides good systematic reviews and taxonomies of these”
Explainable AI: from black box to glass box (2019, background): “A number of classifications of XAI techniques for deep learning models have been proposed (e.g., Hall and Gill 2019; Du et al. 2018; Ribeiro et al. 2016).”
High-resolution Controllable Prostatic Histology Synthesis using StyleGAN (2021, background): “In a review of interpretable models, Du et al. point out that many machine learning models sacrifice accuracy for interpretability (Du et al., 2019).”