{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,2]],"date-time":"2026-04-02T21:49:46Z","timestamp":1775166586962,"version":"3.50.1"},"reference-count":31,"publisher":"MDPI AG","issue":"8","license":[{"start":{"date-parts":[[2025,7,29]],"date-time":"2025-07-29T00:00:00Z","timestamp":1753747200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Entropy"],"abstract":"<jats:p>We propose a novel interpretability framework that integrates instance-wise feature selection with causal reasoning to explain decisions made by black-box image classifiers. Instead of relying on feature importance or mutual information, our method identifies input regions that exert the greatest causal influence on model predictions. Causal influence is formalized using a structural causal model and quantified via a conditional mutual information term. To optimize this objective efficiently, we employ continuous subset sampling and the matrix-based R\u00e9nyi\u2019s \u03b1-order entropy functional. The resulting explanations are compact, semantically meaningful, and causally grounded. Experiments across multiple vision datasets demonstrate that our method outperforms existing baselines in terms of predictive fidelity.<\/jats:p>","DOI":"10.3390\/e27080814","type":"journal-article","created":{"date-parts":[[2025,7,29]],"date-time":"2025-07-29T16:16:10Z","timestamp":1753805770000},"page":"814","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":19,"title":["Causally-Informed Instance-Wise Feature Selection for Explaining Visual Classifiers"],"prefix":"10.3390","volume":"27","author":[{"given":"Li","family":"Tan","sequence":"first","affiliation":[{"name":"Adobe, San Francisco, CA 94103, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2025,7,29]]},"reference":[{"key":"ref_1","unstructured":"Adebayo, J., Gilmer, J., Muelly, M., Goodfellow, I., Hardt, M., and Kim, B. (2018, January 3\u20138). Sanity checks for saliency maps. Proceedings of the Advances in nEural Information Processing Systems, Montreal, QC, Canada."},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Tomsett, R., Harborne, D., Chakraborty, S., Gurram, P., and Preece, A. (2020, January 7\u201312). Sanity checks for saliency metrics. Proceedings of the AAAI Conference on Artificial Intelligence, New York, NY, USA.","DOI":"10.1609\/aaai.v34i04.6064"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Bang, S., Xie, P., Lee, H., Wu, W., and Xing, E. (2021, January 2\u20139). Explaining a black-box by using a deep variational information bottleneck approach. Proceedings of the AAAI Conference on Artificial Intelligence, Virtually.","DOI":"10.1609\/aaai.v35i13.17358"},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Choi, C., Yu, S., Kampffmeyer, M., Salberg, A.B., Handegard, N.O., and Jenssen, R. (2024, January 14\u201319). DIB-X: Formulating Explainability Principles for a Self-Explainable Model Through Information Theoretic Learning. Proceedings of the ICASSP 2024\u20142024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Seoul, Republic of Korea.","DOI":"10.1109\/ICASSP48485.2024.10447094"},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Panda, P., Kancheti, S.S., and Balasubramanian, V.N. (2021, January 20\u201325). Instance-wise causal feature selection for model interpretation. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, Nashville, TN, USA.","DOI":"10.1109\/CVPRW53098.2021.00194"},{"key":"ref_6","unstructured":"Chen, J., Song, L., Wainwright, M., and Jordan, M. (2018, January 10\u201315). Learning to explain: An information-theoretic perspective on model interpretation. Proceedings of the International Conference on Machine Learning (PMLR), Stockholm, Sweden."},{"key":"ref_7","unstructured":"Yoon, J., Jordon, J., and van der Schaar, M. (2019, January 6\u20139). INVASE: Instance-wise Variable Selection using Neural Networks. Proceedings of the International Conference on Learning Representations, New Orleans, LA, USA."},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Pearl, J. (2009). Causality, Cambridge University Press.","DOI":"10.1017\/CBO9780511803161"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"612","DOI":"10.1109\/JPROC.2021.3058954","article-title":"Toward causal representation learning","volume":"109","author":"Locatello","year":"2021","journal-title":"Proc. IEEE"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Caruana, R., Lou, Y., Gehrke, J., Koch, P., Sturm, M., and Elhadad, N. (2015, January 10\u201313). Intelligible models for healthcare: Predicting pneumonia risk and hospital 30-day readmission. Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Sydney, Australia.","DOI":"10.1145\/2783258.2788613"},{"key":"ref_11","unstructured":"Doshi-Velez, F., and Kim, B. (2017). Towards a rigorous science of interpretable machine learning. arXiv."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"17","DOI":"10.1142\/S0219525908001465","article-title":"Information flows in causal networks","volume":"11","author":"Ay","year":"2008","journal-title":"Adv. Complex Syst."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"2324","DOI":"10.1214\/13-AOS1145","article-title":"Quantifying causal influences","volume":"41","author":"Janzing","year":"2013","journal-title":"Ann. Stat."},{"key":"ref_14","first-page":"2960","article-title":"Multivariate Extension of Matrix-Based R\u00e9nyi\u2019s \u03b1-Order Entropy Functional","volume":"42","author":"Yu","year":"2019","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_15","unstructured":"Jang, E., Gu, S., and Poole, B. (2017, January 24\u201326). Categorical Reparametrization with Gumble-Softmax. Proceedings of the International Conference on Learning Representations, Toulon, France."},{"key":"ref_16","unstructured":"Simonyan, K., Vedaldi, A., and Zisserman, A. (2014, January 14\u201316). Deep inside convolutional networks: Visualising image classification models and saliency maps. Proceedings of the International Conference on Learning Representations (ICLR), Banff, Canada, AB."},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., and Batra, D. (2017, January 22\u201329). Grad-cam: Visual explanations from deep networks via gradient-based localization. Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy.","DOI":"10.1109\/ICCV.2017.74"},{"key":"ref_18","unstructured":"Lundberg, S.M., and Lee, S.I. (2017, January 4\u20139). A unified approach to interpreting model predictions. Proceedings of the Advances in Neural Information Processing Systems, Long Beach, CA, USA."},{"key":"ref_19","unstructured":"Tishby, N., Pereira, F.C., and Bialek, W. (1999, January 22\u201324). The information bottleneck method. Proceedings of the 37th Annual Allerton Conference on Communication, Monticello, IL, USA."},{"key":"ref_20","unstructured":"Uendes, B., Yu, S., and Hoogendoorn, M. (2025, January 24\u201328). Start Smart: Leveraging Gradients For Enhancing Mask-based XAI Methods. Proceedings of the Thirteenth International Conference on Learning Representations, Singapore."},{"key":"ref_21","unstructured":"Simoes, F.N., Dastani, M., and van Ommen, T. (2025, January 21\u201325). The Causal Information Bottleneck and Optimal Causal Variable Abstractions. Proceedings of the 41st Conference on Uncertainty in Artificial Intelligence, Rio de Janeiro, Brazil."},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Simoes, F.N., Dastani, M., and van Ommen, T. (2024, January 1\u20133). Fundamental properties of causal entropy and information gain. Proceedings of the Causal Learning and Reasoning (PMLR), Los Angeles, CA, USA.","DOI":"10.1007\/978-3-031-50396-2_12"},{"key":"ref_23","unstructured":"MacKay, D.J. (2003). Information Theory, Inference and Learning Algorithms, Cambridge University Press."},{"key":"ref_24","unstructured":"Belghazi, M.I., Baratin, A., Rajeshwar, S., Ozair, S., Bengio, Y., Courville, A., and Hjelm, D. (2018, January 10\u201315). Mutual information neural estimation. Proceedings of the International Conference on Machine Learning (PMLR), Stockholm, Sweden."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"13066","DOI":"10.1109\/TNNLS.2024.3449419","article-title":"Brainib: Interpretable brain network-based psychiatric diagnosis with graph information bottleneck","volume":"36","author":"Zheng","year":"2024","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"535","DOI":"10.1109\/TIT.2014.2370058","article-title":"Measures of entropy from data using infinitely divisible kernels","volume":"61","author":"Giraldo","year":"2014","journal-title":"IEEE Trans. Inf. Theory"},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"221","DOI":"10.1080\/00029890.2006.11920300","article-title":"Infinitely divisible matrices","volume":"113","author":"Bhatia","year":"2006","journal-title":"Am. Math. Mon."},{"key":"ref_28","unstructured":"Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., and Antiga, L. (2019, January 8\u201314). PyTorch: An Imperative Style, High-Performance Deep Learning Library. Proceedings of the Advances in Neural Information Processing Systems, Vancouver, BC, Canada."},{"key":"ref_29","unstructured":"Xiao, H., Rasul, K., and Vollgraf, R. (2017). Fashion-mnist: A novel image dataset for benchmarking machine learning algorithms. arXiv."},{"key":"ref_30","unstructured":"Krizhevsky, A., and Hinton, G. (2025, May 01). Learning Multiple Layers of Features from Tiny Images. Available online: http:\/\/www.cs.utoronto.ca\/~kriz\/learning-features-2009-TR.pdf."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"6170","DOI":"10.1109\/TSP.2022.3233724","article-title":"Computationally Efficient Approximations for Matrix-Based Renyi\u2019s Entropy","volume":"70","author":"Gong","year":"2023","journal-title":"IEEE Trans. Signal Process."}],"container-title":["Entropy"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1099-4300\/27\/8\/814\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,9]],"date-time":"2025-10-09T18:18:20Z","timestamp":1760033900000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1099-4300\/27\/8\/814"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,7,29]]},"references-count":31,"journal-issue":{"issue":"8","published-online":{"date-parts":[[2025,8]]}},"alternative-id":["e27080814"],"URL":"https:\/\/doi.org\/10.3390\/e27080814","relation":{},"ISSN":["1099-4300"],"issn-type":[{"value":"1099-4300","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,7,29]]}}}