[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117919-en":3,"doc-seo-117919-105":29,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":11,"language":21,"language_code":22,"site_id":23,"html_lang":22,"table_of_contents":24,"faqs":25,"seo_title":26,"seo_description":14,"update_tm":27,"read_time":28},117919,16904993612988,"Olivia Brown","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","Can I Trust the Explanations? Investigating Explainable Machine Learning Methods for Monotonic Models","Explainable machine learning methods have gained strong adoption, yet most approaches target black-box models and may not yield reliable, consistent scientific explanations when domain knowledge is built into the predictor. This study examines monotonic, science-informed machine learning models with three distinct monotonicity notions. Three axioms are proposed to formalize average monotonicity preservation, and attribution quality is assessed. Results indicate Shapley value explanations work well under individual monotonicity, while integrated gradients gives reasonable average explanations under strong pairwise monotonicity.","Can I Trust the Explanations? Investigating Explainable Machine Learning Methods for Monotonic Models  \nDangxing Chen  \nKunshan Duke University  \nKunshan, China  \n[dangxing.chen@dukekunshan.edu.cn](dangxing.chen@dukekunshan.edu.cn)  \narXiv :2309 . 13246v1 [ cs .LG] 23 Sep 2023  \nABSTRACT  \nIn recent years, explainable machine learning methods have been very successful. Despite their success, most explainable machine learning methods are applied to black-box models without any domain knowledge. By incorporating domain knowledge, scienceinformed machine learning models have demonstrated better generalization and interpretation. But do we obtain consistent scientific explanations if we apply explainable machine learning methods to science-informed machine learning models? This question is addressed in the context of monotonic models that exhibit three different types of monotonicity. To demonstrate monotonicity, we propose three axioms. Accordingly, this study shows that when only individual monotonicity is involved, the baseline Shapley value provides good explanations; however, when strong pairwise monotonicity is involved, the Integrated gradients method provides reasonable explanations on average.  \nKEYWORDS  \nexplainable machine learning, monotonicity, fairness, neural networks  \n1 INTRODUCTION  \nIn recent decades, machine learning (ML) models have achieved many successes. In comparison with traditional methods, machine learning models are often capable of increasing accuracy at the expense of black-box functionality. The importance of model explanation is particularly important for highly regulated industries such as the financial sector [1] . As an example, the Consumer Financial Protection Bureau (CFPB)confirmed that anti-discrimination law requires companies to provide detailed explanations when denying an application for credit when using machine learning methods 1. In response to the growing regulatory requirements, researchers are investigating explainable machine learning methods.  \nExplainable machine learning (XML) methods have been successfully used in the past to achieve great success in machine learning. Popular used methods include SHapley Additive exPlanations (SHAP)[15], Local Interpretable Model-Agnostic Explanations(LIME)  \n[19], Integrated Gradients (IG) [22], Anchors [20], and Sensitivitybased methods [11] . As a result of these methods, we have gained a better understanding of how ML models function [3, 4, 12] .  \nIn this paper, we address the attributionproblem, which involves allocating the prediction score of a model for a given input to its base features. The attribution to a base feature can be understood as the importance of the feature in the prediction. Credit scoring, for instance, can utilize attribution to understand how each feature  \n1[https://www.consumer](https://www.consumer)fi[nance.gov/about-us/newsroom/cfpb-acts-to-protect-the](nance.gov/about-us/newsroom/cfpb-acts-to-protect-the)public-from-black-box-credit-models-using-complex-algorithms/  \ncontributes to the credit score. The SHAP and IG have been successfully applied to the attribution problem. Further, their use has been demonstrated to comply with a number of theoretical properties, as discussed in [15, 16, 21, 22]. While extensive analyses have been conducted, the majority of results have been based on blackbox machine learning models without domain knowledge.  \nModeling and ensuring conceptual soundness require domain knowledge: the true model should be consistent with the underlying theories. A number of studies have demonstrated that physicsinformed machine learning (PIML) [9, 13] improved black-box machine learning models in terms of interpretation and accuracy by enforcing conservation laws, for example. Finance and other applications often require monotonicity. A person’s credit score should be decreased when there is one more past due balance on the account, for example. It is possible to achieve better generalization and inte","cbCaigofexADICnn","https://ap.wps.com/l/cbCaigofexADICnn","pdf",209513,1,"English","en",105,"# Abstract\n# Introduction\n# Problem Setting and Attribution\n## Monotonicity and Fairness in Science-Informed Models\n# Contributions and Proposed Axioms\n## Average Preservation of Monotonicity Types\n# Explanation Methods Compared","[{\"question\":\"Why is domain knowledge important for explainable machine learning?\",\"answer\":\"Domain knowledge can make the learned model align with underlying scientific or theoretical principles, improving both interpretation and generalization compared with purely black-box explanations.\"},{\"question\":\"What types of monotonicity are studied in this paper?\",\"answer\":\"The paper considers three different types of monotonicity and proposes axioms to evaluate average preservation of these monotonic properties in attribution explanations.\"},{\"question\":\"Which attribution method performs well under individual versus pairwise monotonicity?\",\"answer\":\"When only individual monotonicity is involved, the baseline Shapley value provides good explanations; when strong pairwise monotonicity is involved, integrated gradients provides reasonable explanations on average.\"}]","Can I Trust the Explanations? Investigating Explainable Machine Learning Methods for Monotonic Models | PDF",1785680382,20,{"code":4,"msg":30,"data":31},"ok",{"site_id":23,"language":22,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"can-i-trust-the-explanations-investigating-explainable-machine-learning-methods-for-monotonic-models","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/can-i-trust-the-explanations-investigating-explainable-machine-learning-methods-for-monotonic-models/117919/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":22,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"Why is domain knowledge important for explainable machine learning?","Question",{"text":74,"@type":75},"Domain knowledge can make the learned model align with underlying scientific or theoretical principles, improving both interpretation and generalization compared with purely black-box explanations.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"What types of monotonicity are studied in this paper?",{"text":79,"@type":75},"The paper considers three different types of monotonicity and proposes axioms to evaluate average preservation of these monotonic properties in attribution explanations.",{"name":81,"@type":72,"acceptedAnswer":82},"Which attribution method performs well under individual versus pairwise monotonicity?",{"text":83,"@type":75},"When only individual monotonicity is involved, the baseline Shapley value provides good explanations; when strong pairwise monotonicity is involved, integrated gradients provides reasonable explanations on average.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":23},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":28,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":28,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":28,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]