--- title: Consistent security model deployment with FPR calibration description: Cisco AI Defense open-sources an FPR (false-positive rate) calibration method for security detection models. The core problem: when a model is retrained and... canonical: https://daily.dev/posts/consistent-security-model-deployment-with-fpr-calibration-jhnpk6vcj twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Consistent security model deployment with FPR calibration | daily.dev og:description: Cisco AI Defense open-sources an FPR (false-positive rate) calibration method for security detection models. The core problem: when a model is retrained and... og:url: https://daily.dev/posts/consistent-security-model-deployment-with-fpr-calibration-jhnpk6vcj og:image: https://api.daily.dev/og/posts/jhnpk6VCj.png og:image:alt: Consistent security model deployment with FPR calibration og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Consistent security model deployment with FPR calibration **[Cisco](https://daily.dev/sources/cisco)** · 5 min read · 0 upvotes · 0 comments ## Summary Cisco AI Defense open-sources an FPR (false-positive rate) calibration method for security detection models. The core problem: when a model is retrained and redeployed, raw score thresholds shift, silently changing how aggressively it blocks traffic. FPR calibration maps raw model scores to a stable score contract tied to benign traffic rates, so policy thresholds retain their meaning across model versions. The calibrator is a bounded sklearn artifact fitted on benign scores offline. The post covers why FPR is preferable to probability calibration for security settings (benign traffic is stable; attack distributions are not), how much benign data is needed for reliable calibration at various operating points, and validation on the Credit Card Fraud Detection benchmark. Code is available on GitHub with a preprint on arXiv. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Your Model’s 0.9 Probably Isn’t 90%](https://daily.dev/posts/your-model-s-0-9-probably-isn-t-90--fhtgyspvh) · Medium · 0 upvotes · 0 comments - [The AI Model Confidence Trap](https://daily.dev/posts/the-ai-model-confidence-trap-93to0xhgy) · Towards Data Science · 5 upvotes · 1 comments --- Tags: [#security](https://daily.dev/tags/security), [#machine-learning](https://daily.dev/tags/machine-learning), [#scikit](https://daily.dev/tags/scikit) [View this post on daily.dev](https://daily.dev/posts/consistent-security-model-deployment-with-fpr-calibration-jhnpk6vcj) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Consistent security model deployment with FPR calibration","url":"https://daily.dev/posts/consistent-security-model-deployment-with-fpr-calibration-jhnpk6vcj","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/consistent-security-model-deployment-with-fpr-calibration-jhnpk6vcj"},"datePublished":"2026-07-13T21:24:40.732Z","dateModified":"2026-07-13T21:25:45.246Z","description":"Cisco AI Defense open-sources an FPR (false-positive rate) calibration method for security detection models. The core problem: when a model is retrained and...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/9c17efe3338e59a91fe4f52838cdba03?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/9c17efe3338e59a91fe4f52838cdba03?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Cisco","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Cisco","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/68b8a8be17184235a5067d7c84162f77","url":"https://daily.dev/sources/cisco"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/consistent-security-model-deployment-with-fpr-calibration-jhnpk6vcj","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"security,machine-learning,scikit","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Cisco","item":"https://daily.dev/sources/cisco"},{"@type":"ListItem","position":3,"name":"Consistent security model deployment with FPR calibration"}]} ```