ABSTRACT
Counterfactuals, serving as one of the emerging type of model interpretations, have recently received attention from both researchers and practitioners. Counterfactual explanations formalize the exploration of "what-if'' scenarios, and are an instance of example-based reasoning using a set of hypothetical data samples. Counterfactuals essentially show how the model decision alters with input perturbations. Existing methods for generating counterfactuals are mainly algorithm-based, which are time-inefficient and assume the same counterfactual universe for different queries. To address these limitations, we propose a Model-based Counterfactual Synthesizer (MCS) framework for interpreting machine learning models. We first analyze the model-based counterfactual process and construct a base synthesizer using a conditional generative adversarial net (CGAN). To better approximate the counterfactual universe for those rare queries, we novelly employ the umbrella sampling technique to conduct the MCS framework training. Besides, we also enhance the MCS framework by incorporating the causal dependence among attributes with model inductive bias, and validate its design correctness from the causality identification perspective. Experimental results on several datasets demonstrate the effectiveness as well as efficiency of our proposed MCS framework, and verify the advantages compared with other alternatives.
- Peter W Battaglia, Jessica B Hamrick, Victor Bapst, et al. 2018. Relational inductive biases, deep learning, and graph networks. arXiv:1806.01261 (2018).Google Scholar
- Christopher M Bishop. 2006. Pattern recognition and machine learning. springer.Google Scholar
- Tim Brennan and William L Oliver. 2013. Emergence of machine learning techniques in criminology: implications of complexity in our data and in research questions. Criminology & Pub. Pol'y, Vol. 12 (2013), 551.Google ScholarCross Ref
- Chaofan Chen, Oscar Li, Daniel Tao, Alina Barnett, Cynthia Rudin, and Jonathan K Su. 2019. This looks like that: deep learning for interpretable image recognition. In Advances in Neural Information Processing Systems (NeurIPS). 8928--8939.Google Scholar
- Amit Dhurandhar, Pin-Yu Chen, et al. 2018. Explanations based on the missing: Towards contrastive explanations with pertinent negatives. In NeurIPS. 592--603.Google Scholar
- Mengnan Du, Ninghao Liu, and Xia Hu. 2019. Techniques for interpretable machine learning. Commun. ACM, Vol. 63, 1 (2019), 68--77.Google ScholarDigital Library
- José AR Fonollosa. 2019. Conditional distribution variability measures for causality detection. In Cause Effect Pairs in Machine Learning. Springer, 339--347.Google Scholar
- Amirata Ghorbani, James Wexler, James Zou, and Been Kim. 2019. Towards automatic concept-based explanations. arXiv preprint arXiv:1902.03129 (2019).Google Scholar
- Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, et al. 2014. Generative adversarial nets. In NeurIPS. 2672--2680.Google Scholar
- Yash Goyal, Amir Feder, Uri Shalit, and Been Kim. 2019 a. Explaining classifiers with causal concept effect (cace). arXiv:1907.07165 (2019).Google Scholar
- Yash Goyal, Ziyan Wu, Jan Ernst, et al. 2019 b. Counterfactual Visual Explanations. In International Conference on Machine Learning (ICML). 2376--2384.Google Scholar
- Thomas L Griffiths, Nick Chater, Charles Kemp, Amy Perfors, and Joshua B Tenenbaum. 2010. Probabilistic models of cognition: Exploring representations and inductive biases. Trends in cognitive sciences, Vol. 14, 8 (2010), 357--364.Google Scholar
- Ned Hall. 2007. Structural equations and causation. Philosophical Studies, Vol. 132, 1 (2007), 109--136.Google ScholarCross Ref
- Patrik Hoyer, Dominik Janzing, Joris M Mooij, Jonas Peters, et al. 2008. Nonlinear causal discovery with additive noise models. NeurIPS, Vol. 21 (2008), 689--696.Google Scholar
- Eric Jang, Shixiang Gu, and Ben Poole. 2016. Categorical reparameterization with gumbel-softmax. arXiv:1611.01144 (2016).Google Scholar
- Shalmali Joshi, Oluwasanmi Koyejo, Been Kim, et al. 2018. xGEMs: Generating examplars to explain black-box models. arXiv:1806.08867 (2018).Google Scholar
- Diviyan Kalainathan, Olivier Goudet, and Ritik Dutta. 2020. Causal Discovery Toolbox: Uncovering causal relationships in Python. JMLR, Vol. 21, 37 (2020), 1--5.Google Scholar
- Amir-Hossein Karimi, Gilles Barthe, Bernhard Schölkopf, and Isabel Valera. 2020. A survey of algorithmic recourse: definitions, formulations, solutions, and prospects. arXiv:2010.04050 (2020).Google Scholar
- Johannes Kästner. 2011. Umbrella sampling. Wiley Interdisciplinary Reviews: Computational Molecular Science, Vol. 1, 6 (2011), 932--942.Google ScholarCross Ref
- Been Kim, Rajiv Khanna, and Oluwasanmi O Koyejo. 2016. Examples are not enough, learn to criticize! criticism for interpretability. In NeurIPS. 2280--2288.Google Scholar
- Been Kim, Martin Wattenberg, Justin Gilmer, Carrie Cai, et al. 2018. Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV). In ICML. 2668--2677.Google Scholar
- Pang Wei Koh and Percy Liang. 2017. Understanding black-box predictions via influence functions. In ICML. 1885--1894.Google Scholar
- Geert Litjens, Thijs Kooi, Babak Ehteshami Bejnordi, Arnaud Arindra Adiyoso Setio, Francesco Ciompi, et al. 2017. A survey on deep learning in medical image analysis. Medical image analysis, Vol. 42 (2017), 60--88.Google Scholar
- Scott M Lundberg and Su-In Lee. 2017. A unified approach to interpreting model predictions. In NeurIPS. 4765--4774.Google Scholar
- Divyat Mahajan, Chenhao Tan, and Amit Sharma. 2019. Preserving causal constraints in counterfactual explanations for machine learning classifiers. arXiv:1912.03277 (2019).Google Scholar
- JL McClelland. 1992. The interaction of nature and nurture in development: A parallel distributed processing perspective (Parallel Distributed Processing and Cognitive Neuroscience PDP. CNS. 92.6).Google Scholar
- Mehdi Mirza and Simon Osindero. 2014. Conditional generative adversarial nets. arXiv:1411.1784 (2014).Google Scholar
- Tom M Mitchell. 1980. The need for biases in learning generalizations. Department of Computer Science, Laboratory for Computer Science Research.Google Scholar
- Jonathan Moore, Nils Hammerla, and Chris Watkins. 2019. Explaining deep learning models with constrained adversarial examples. In Pacific Rim International Conference on Artificial Intelligence. Springer, 43--56.Google ScholarDigital Library
- Ramaravind K Mothilal, Amit Sharma, and Chenhao Tan. 2020. Explaining machine learning classifiers through diverse counterfactual explanations. In Proceedings on Fairness, Accountability, and Transparency (FAccT). 607--617.Google ScholarDigital Library
- Noseong Park, Mahmoud Mohammadi, Kshitij Gorde, Sushil Jajodia, Hongkyu Park, and Youngmin Kim. 2018. Data synthesis based on generative adversarial networks. arXiv:1806.03384 (2018).Google Scholar
- Martin Pawelczyk, Klaus Broelemann, and Gjergji Kasneci. 2020. Learning Model-Agnostic Counterfactual Explanations for Tabular Data. In Proceedings of The Web Conference 2020. 3126--3132.Google ScholarDigital Library
- Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. 2016. "Why should i trust you?" Explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD conference on knowledge discovery and data mining. 1135--1144.Google ScholarDigital Library
- Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. 2018. Anchors: High-precision model-agnostic explanations. In 32nd AAAI on Artificial Intelligence.Google Scholar
- Edwina L Rissland. 1991. Example-based reasoning. Informal reasoning in education (1991), 187--208.Google Scholar
- Ramprasaath R Selvaraju, Michael Cogswell, et al. 2017. Grad-Cam: Visual explanations from deep networks via gradient-based localization. In Proceedings of the IEEE international conference on computer vision (ICCV). 618--626.Google ScholarCross Ref
- Mukund Sundararajan, Ankur Taly, and Qiqi Yan. 2017. Axiomatic attribution for deep networks. In Proceedings of the 34th International Conference on Machine Learning-Volume 70. JMLR. org, 3319--3328.Google ScholarDigital Library
- Tom Vermeire and David Martens. 2020. Explainable Image Classification with Evidence Counterfactual. arXiv:2004.07511 (2020).Google Scholar
- Sandra Wachter et al. 2017. Counterfactual explanations without opening the black box: Automated decisions and the GDPR. Harv. JL & Tech., Vol. 31 (2017), 841.Google Scholar
- Adam White and Artur d'Avila Garcez. 2019. Measurable counterfactual local explanations for any classifier. arXiv preprint arXiv:1908.03020 (2019).Google Scholar
- Lei Xu, Maria Skoularidou, Alfredo Cuesta-Infante, and Kalyan Veeramachaneni. 2019. Modeling tabular data using conditional gan. In NeurIPS. 7335--7345.Google Scholar
- Fan Yang, Ninghao Liu, Mengnan Du, et al. 2021. Generative Counterfactuals for Neural Networks via Attribute-Informed Perturbation. arXiv:2101.06930 (2021).Google Scholar
- Fan Yang, Ninghao Liu, Mengnan Du, Kaixiong Zhou, Shuiwang Ji, and Xia Hu. 2020. Deep Neural Networks with Knowledge Instillation. In SDM. 370--378.Google Scholar
- Chih-Kuan Yeh, Been Kim, et al. 2020. On Completeness-aware Concept-Based Explanations in Deep Neural Networks. NeurIPS (2020).Google Scholar
- Bolei Zhou, Yiyou Sun, David Bau, and Antonio Torralba. 2018. Interpretable basis decomposition for visual explanation. In ECCV. 119--134.Google Scholar
Index Terms
- Model-Based Counterfactual Synthesizer for Interpretation
Recommendations
Counterfactual learning in customer churn prediction under class imbalance
ICBDT '23: Proceedings of the 2023 6th International Conference on Big Data TechnologiesNowadays, in churn prediction, many decision-makers are trying to obtain knowledge from models through interpretation techniques due to the non-transparency of black-box. In these techniques, the counterfactual explanation generated by the ...
Ontology-Based Inference for Causal Explanation
Knowledge Science, Engineering and ManagementAbstractWe define an inference system to capture explanations based on causal statements, using an ontology in the form of an IS-A hierarchy. We first introduce a simple logical language which makes it possible to express that a fact causes another fact ...
Explainable Agency in Intelligent Agents: Doctoral Consortium
AAMAS '19: Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent SystemsExplainability of intelligent agents has gained attention in recent years with their widespread utilization in society. Most work in Explainable AI (XAI) pay little attention to the humans that interact with these models, which risks resulting in ...
Comments