{"id":917364,"date":"2023-02-06T17:19:42","date_gmt":"2023-02-07T01:19:42","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-project&#038;p=917364"},"modified":"2023-06-13T08:20:19","modified_gmt":"2023-06-13T15:20:19","slug":"tools-for-managing-and-ideating-responsible-ai-mitigations","status":"publish","type":"msr-project","link":"https:\/\/www.microsoft.com\/en-us\/research\/project\/tools-for-managing-and-ideating-responsible-ai-mitigations\/","title":{"rendered":"Tools for Managing and Ideating Responsible AI Mitigations"},"content":{"rendered":"<section class=\"mb-3 moray-highlight\">\n\t<div class=\"card-img-overlay mx-lg-0\">\n\t\t<div class=\"card-background  has-background-plum card-background--full-bleed\">\n\t\t\t<img loading=\"lazy\" decoding=\"async\" width=\"2000\" height=\"728\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Ideating-Responsible-AI-Mitigations-Background.png\" class=\"attachment-full size-full\" alt=\"General Header Image Responsible AI\" style=\"\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Ideating-Responsible-AI-Mitigations-Background.png 2000w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Ideating-Responsible-AI-Mitigations-Background-300x109.png 300w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Ideating-Responsible-AI-Mitigations-Background-1024x373.png 1024w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Ideating-Responsible-AI-Mitigations-Background-768x280.png 768w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Ideating-Responsible-AI-Mitigations-Background-1536x559.png 1536w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Ideating-Responsible-AI-Mitigations-Background-240x87.png 240w\" sizes=\"(max-width: 2000px) 100vw, 2000px\" \/>\t\t<\/div>\n\t\t<!-- Foreground -->\n\t\t<div class=\"card-foreground d-flex mt-md-n5 my-lg-5 px-g px-lg-0\">\n\t\t\t<!-- Container -->\n\t\t\t<div class=\"container d-flex mt-md-n5 my-lg-5 \">\n\t\t\t\t<!-- Card wrapper -->\n\t\t\t\t<div class=\"w-100 w-lg-col-5\">\n\t\t\t\t\t<!-- Card -->\n\t\t\t\t\t<div class=\"card material-md-card py-5 px-md-5\">\n\t\t\t\t\t\t<div class=\"card-body \">\n\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\n\n<h2 class=\"wp-block-heading\" id=\"ideating-responsible-ai-mitigations\">Ideating Responsible AI Mitigations<\/h2>\n\n\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t<\/div>\n\t<\/div>\n<\/section>\n\n\n\n\n\n<p>News: Our slides from the FAccT Tutorial on Responsible AI Toolbox are available <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/06\/responsible_ai_toolbox_facct_tutorial_2023.pdf\">here<\/a>.<\/p>\n\n\n\n<p>ML algorithms and systems are often prone to severe bias and highly consequential failure modes that are not well understood. This project advances the methods, tools, and infrastructure for debugging and mitigating these failure modes so practitioners may act on them before deploying ML systems in the real world. The project is part of <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/github.com\/microsoft\/responsible-ai-toolbox\">Responsible AI Toolbox<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>, a larger collaborative effort between Microsoft Research, AETHER, and Azure Machine Learning for integrating and building development tools for responsible AI.<\/p>\n\n\n\n<p>The goal of this project is two-fold:&nbsp;<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>Building tools that enable ML engineers to identify, diagnose, and mitigate problems quickly and systematically.&nbsp;<\/li>\n\n\n\n<li>Conducting research that supports the above processes by better understanding and improving algorithmic robustness and failure explainability for different model architectures and data types.<\/li>\n<\/ol>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"576\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-1024x576.jpg\" alt=\"Flowchart showing how responsible AI tools are used together for targeted debugging of machine learning models: the Responsible AI Dashboard for the identification of failures; followed by the Responsible AI Dashboard and Mitigations Library for the diagnosis of failures; then the Responsible AI Mitigations Library for mitigating failures; and lastly the Responsible AI Tracker for tracking, comparing, and validating mitigation techniques from which an arrow points back to the identification phase of the cycle to indicate the repetition of the process as models and data continue to evolve during the ML lifecycle.\" class=\"wp-image-917367\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-1024x576.jpg 1024w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-300x169.jpg 300w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-768x432.jpg 768w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-1066x600.jpg 1066w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-655x368.jpg 655w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-343x193.jpg 343w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-240x135.jpg 240w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-640x360.jpg 640w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-960x540.jpg 960w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox-1280x720.jpg 1280w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2023\/02\/Responsible-AI-Toolbox.jpg 1400w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p><strong>Recent releases<\/strong>:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/github.com\/microsoft\/responsible-ai-toolbox-tracker\"><strong>Responsible AI Tracker<\/strong> <span class=\"sr-only\"> (opens in new tab)<\/span><\/a>A Jupyter Lab extension for managing, tracking, and comparing different RAI mitigation experiments. The goal is to accelerate improvement iterations for ML practitioners by enabling them to experiment and compare mitigation results quickly.&nbsp;<\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/github.com\/microsoft\/responsible-ai-toolbox-mitigations\" target=\"_blank\" rel=\"noreferrer noopener\"><strong>Responsible AI Mitigations library<\/strong><span class=\"sr-only\"> (opens in new tab)<\/span><\/a> is the ML backend support for targeted mitigation steps that can be used in the Responsible AI Tracker or in any other RAI tool. The designed functionalities guide model improvement by targeting mitigations to errors that affect particular data cohorts, with the goal of reducing performance discrepancies across cohorts.&nbsp;<\/li>\n<\/ul>\n\n\n\n<p><strong>Past releases<\/strong>: In collaboration with Azure Machine Learning, AETHER, and the Mixed Reality Group, we have built the following Responsible AI tools:&nbsp;<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"http:\/\/erroranalysis.ai\/\" target=\"_blank\" rel=\"noreferrer noopener\"><strong>Error Analysis<\/strong><span class=\"sr-only\"> (opens in new tab)<\/span><\/a> for identifying and diagnosing errors of ML models and systems.&nbsp;<\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/responsibleaitoolbox.ai\/\" target=\"_blank\" rel=\"noreferrer noopener\"><strong>Responsible AI Dashboard<\/strong><\/a> as a one-stop shop dashboard for integrating together several Responsible AI tools on error analysis, interpretability, causality, fairness, and decision making. The dashboard builds upon other RAI offerings at Microsoft such as <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"http:\/\/erroranalysis.ai\/\" target=\"_blank\" rel=\"noreferrer noopener\">Error Analysis<\/a>, <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"http:\/\/fairlearn.org\/\" target=\"_blank\" rel=\"noreferrer noopener\">Fairlearn<\/a>, <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"http:\/\/interpret.ml\/\" target=\"_blank\" rel=\"noreferrer noopener\">InterpretML<\/a>, and <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/project\/econml\/\" target=\"_blank\" rel=\"noreferrer noopener\">EconML<\/a>.&nbsp;<\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/github.com\/microsoft\/backwardcompatibilityML\" target=\"_blank\" rel=\"noreferrer noopener\"><strong>BackwardCompatibilityML<\/strong><span class=\"sr-only\"> (opens in new tab)<\/span><\/a> for training ML models that do not regress and do not introduce new errors. The tool also provides visualizations for model comparison.&nbsp;<\/li>\n<\/ul>\n\n\n","protected":false},"excerpt":{"rendered":"<p>News: Our slides from the FAccT Tutorial on Responsible AI Toolbox are available here. ML algorithms and systems are often prone to severe bias and highly consequential failure modes that are not well understood. This project advances the methods, tools, and infrastructure for debugging and mitigating these failure modes so practitioners may act on them [&hellip;]<\/p>\n","protected":false},"featured_media":920094,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"footnotes":""},"research-area":[13556],"msr-locale":[268875],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-917364","msr-project","type-msr-project","status-publish","has-post-thumbnail","hentry","msr-research-area-artificial-intelligence","msr-locale-en_us","msr-archive-status-active"],"msr_project_start":"","related-publications":[561399,669123,752377,493274],"related-downloads":[],"related-videos":[],"related-groups":[],"related-events":[],"related-opportunities":[],"related-posts":[],"related-articles":[],"tab-content":[],"slides":[],"related-researchers":[{"type":"user_nicename","display_name":"Marah Abdin","user_id":39657,"people_section":"Related people","alias":"maabdin"},{"type":"guest","display_name":"Jingya Chen","user_id":767776,"people_section":"Related people","alias":""},{"type":"user_nicename","display_name":"Mark Encarnaci&oacute;n","user_id":32814,"people_section":"Related people","alias":"markenc"},{"type":"user_nicename","display_name":"Rahee Ghosh Peshawaria","user_id":36864,"people_section":"Related people","alias":"raghosh"},{"type":"user_nicename","display_name":"Eric Horvitz","user_id":32033,"people_section":"Related people","alias":"horvitz"},{"type":"user_nicename","display_name":"Ece Kamar","user_id":31710,"people_section":"Related people","alias":"eckamar"},{"type":"user_nicename","display_name":"Matheus Mendon\u00e7a","user_id":41940,"people_section":"Related people","alias":"mmendonca"},{"type":"user_nicename","display_name":"Besmira Nushi","user_id":36975,"people_section":"Related people","alias":"benushi"},{"type":"user_nicename","display_name":"ThuVan Pham","user_id":34038,"people_section":"Related people","alias":"thuvanp"},{"type":"user_nicename","display_name":"Dany Rouhana","user_id":31540,"people_section":"Related people","alias":"danyr"}],"msr_research_lab":[199565],"msr_impact_theme":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/917364"}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-project"}],"version-history":[{"count":12,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/917364\/revisions"}],"predecessor-version":[{"id":948609,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/917364\/revisions\/948609"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media\/920094"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=917364"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=917364"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=917364"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=917364"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=917364"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}