[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-290913-105":59,"doc-detail-290913-en":123},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":116,"head_meta":118,"extra_data":120,"updated_unix":122},105,"en","privacy-preserving-model-and-preprocessing-verification-for-machine-learning","Privacy-Preserving Model and Preprocessing Verification for Machine Learning","","This paper presents a framework for privacy-preserving verification of machine learning models trained on sensitive data. By combining Local Differential Privacy (LDP) with model explanations from LIME and SHAP, the approach verifies correct training behavior while maintaining individual privacy. It supports two tasks: binary classification to confirm that a target model used the right preprocessing, and multiclass classification to pinpoint specific preprocessing mistakes. Experiments on Diabetes, Adult, and Student Record show strong binary performance and comparable multiclass detection, with useful robustness to noise levels.",{"@graph":69,"@context":115},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/privacy-preserving-model-and-preprocessing-verification-for-machine-learning/290913/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/privacy-preserving-model-and-preprocessing-verification-for-machine-learning/290913.png","ImageObject",300,407,{"name":92,"@type":93},"Eden","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-27","2026-09-17",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",16,{"@type":107,"mainEntity":108},"FAQPage",[109],{"name":110,"@type":111,"acceptedAnswer":112},"Which preprocessing-related tasks does the framework support, and how are they evaluated?","Question",{"text":113,"@type":114},"It supports binary classification to confirm correct preprocessing and multiclass classification to identify particular preprocessing mistakes. Evaluation on Diabetes, Adult, and Student Record compares performance across datasets and noise levels, showing strong binary effectiveness and comparable multiclass results.","Answer","https://schema.org",{"og:url":83,"og:type":117,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":119,"canonical":83},"index,follow",{"doc_id":121,"site_id":62},290913,1789642944,{"code":4,"msg":5,"data":124},{"doc_id":121,"user_id":125,"nickname":92,"user_avatar":126,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":127,"file_id":128,"file_url":129,"file_type":130,"file_size":131,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":132,"language":133,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":134,"faqs":135,"seo_title":136,"seo_description":67,"update_tm":122,"read_time":137,"preview_page_count":4},1374391974468,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","Privacy-Preserving Model and Preprocessing Verification for Machine Learning  \nWenbiao Li  \nCase Western Reserve University  \nAnisa Halimi IBM Research  \nXiaoqian Jiang UThealth  \nJaideep Vaidya Rutgers University  \nErman Ayday  \nCase Western Reserve University  \narXiv :2501 .08236v1 [ cs .LG] 14 Jan 2025  \nAbstract  \nThis paper presents a framework for privacy-preserving verification of machine learning models, focusing on models trained on sensitive data. Integrating Local Differential Privacy (LDP) with model explanations from LIME and SHAP, our framework enables robust verification without compromising individual privacy. It addresses two key tasks: binary classification, to verify if a target model was trained correctly by applying the appropriate preprocessing steps, and multiclass classification, to identify specific preprocessing errors. Evaluations on three real-world datasets—Diabetes, Adult, and Student Record—demonstrate that while the ML-based approach is particularly effective in binary tasks, the thresholdbased method performs comparably in multi-class tasks. Results indicate that although verification accuracy varies across datasets and noise levels, the framework provides effective detection of preprocessing errors, strong privacy guarantees, and practical applicability for safeguarding sensitive data.  \n1 Introduction  \nMachine learning has revolutionized various industries by enabling systems to learn from data and make intelligent decisions [7, 20, 37, 39] . From healthcare diagnostics to financial forecasting, the impact of machine learning is profound and widespread. However, the success of machine learning models hinges not only on the algorithms themselves but also on the quality of the data they are trained on.  \nA crucial step in ensuring data quality is preprocessing [1, 11, 19] . Preprocessing is the stage where raw data is transformed into a clean and usable format. This step includes a variety of tasks such as handling missing values, normalizing data, encoding categorical variables, and removing duplicates. The primary purpose of preprocessing is to  \nenhance the performance of the machine learning model by ensuring that the input data is accurate, consistent, and lacks errors.  \nThe preprocessing stage is susceptible to variations, which can significantly impact the resulting model. For instance,  \nfailing to normalize numerical data might cause the model tobe dominated by attributes with larger magnitudes, leading to biased predictions. Similarly, overlooking the removal of duplicates can skew results. These variations can occur due to factors such as different domain expertise, interpretations, or miscommunication among team members handling the data pipeline. In this work, we consider 6 frequently used preprocessing steps in total, two of them are required before training a model and the remaining four are additional work and decided by the user whether the following study needs one. More details are in 3.4 When improper preprocessing choices are made, models may become less reliable, even if the overall performance metrics appear unaffected noticeably. In critical applications, such as healthcare or finance, these inconsistencies can have serious consequences. Thus, before a model is shared with other users or deployed as a service, it is essential to verify that it has been correctly trained and functions as intended. This verification process helps prevent the propagation of issues that could arise from improper preprocessing approaches, ultimately protecting end-users and the broader systems that rely on these models.  \nHowever, verifying a machine learning model is challenging, especially when it comes to confirming the properness of its preprocessing steps, as the personalized and datasetdependent nature of preprocessing makes improperness difficult to detect. Although some of these improperness may not cause a significant change in overall accuracy, they can still introduce biases, red","cbCaipSQOLBBy0hN","https://ap.wps.com/l/cbCaipSQOLBBy0hN","pdf",826089,17,"English","# Introduction\n## Data quality and preprocessing\n## Challenges of verifying preprocessing correctness\n## Using prediction explanations for verification\n## Privacy-preserving verification with differential privacy","[{\"question\":\"Which preprocessing-related tasks does the framework support, and how are they evaluated?\",\"answer\":\"It supports binary classification to confirm correct preprocessing and multiclass classification to identify particular preprocessing mistakes. Evaluation on Diabetes, Adult, and Student Record compares performance across datasets and noise levels, showing strong binary effectiveness and comparable multiclass results.\"}]","Privacy-Preserving Model and Preprocessing Verification for Machine Learning | PDF",43]