{"id":899655,"date":"2022-11-19T17:47:49","date_gmt":"2022-11-20T01:47:49","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/"},"modified":"2024-08-27T14:10:24","modified_gmt":"2024-08-27T21:10:24","slug":"uglad-recover-conditional-independence-graphs","status":"publish","type":"msr-research-item","link":"https:\/\/www.microsoft.com\/en-us\/research\/publication\/uglad-recover-conditional-independence-graphs\/","title":{"rendered":"uGLAD: A Deep Learning Model to Recover Conditional Independence Graphs"},"content":{"rendered":"<p>Probabilistic Graphical Models are generative models of complex systems. They rely on conditional independence assumptions between variables to learn sparse representations which can be visualized in a form of a graph. Such models are used for domain exploration and structure discovery in poorly understood domains. This work introduces a novel technique to perform sparse graph recovery by optimizing deep unrolled networks. Assuming that the input data comes from an underlying multivariate Gaussian distribution, we apply a deep model on that outputs the precision matrix. Then, the partial correlation matrix is calculated which can also be interpreted as providing a list of conditional independence assertions holding in the input distribution. Our model, uGLAD, builds upon and extends the state-of-the-art model GLAD to the unsupervised setting. The key benefits of our model are (1) uGLAD automatically optimizes sparsity-related regularization parameters leading to better performance than existing algorithms. (2) We introduce multi-task learning based `consensus&#8217; strategy for robust handling of missing data in an unsupervised setting. We evaluate performance on synthetic Gaussian, non-Gaussian data generated from Gene Regulatory Networks, and present case studies in anaerobic digestion and infant mortality.<\/p>\n<p>Software: <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/github.com\/Harshs27\/uGLAD\">uGLAD<\/a><\/p>\n<p>Additional discussions: <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/video\/lightning-talk-a-deep-learning-approach-to-recover-conditional-independence-graphs\/\">Neurips Talk<\/a>, <a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/www.harshshrivastava.com\/\">Tech Blog<\/a><\/p>\n<div id=\"attachment_1080183\" style=\"width: 209px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" aria-describedby=\"caption-attachment-1080183\" class=\"wp-image-1080183 size-large\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prodnew\/2022\/11\/uglad_architecture-199x1024.png\" alt=\"uGLAD deep unrolled architecture\" width=\"199\" height=\"1024\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/uglad_architecture-199x1024.png 199w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/uglad_architecture-35x180.png 35w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/uglad_architecture.png 255w\" sizes=\"(max-width: 199px) 100vw, 199px\" \/><p id=\"caption-attachment-1080183\" class=\"wp-caption-text\">uGLAD&#8217;s deep unrolled architecture<\/p><\/div>\n<div id=\"attachment_1080189\" style=\"width: 310px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" aria-describedby=\"caption-attachment-1080189\" class=\"wp-image-1080189 size-medium\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prodnew\/2022\/11\/nn-architecture1-300x95.png\" alt=\"text\" width=\"300\" height=\"95\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/nn-architecture1-300x95.png 300w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/nn-architecture1-1024x323.png 1024w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/nn-architecture1-768x242.png 768w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/nn-architecture1-1536x484.png 1536w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/nn-architecture1-240x76.png 240w, https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/nn-architecture1.png 1583w\" sizes=\"(max-width: 300px) 100vw, 300px\" \/><p id=\"caption-attachment-1080189\" class=\"wp-caption-text\">Parameterized &#8220;learnable&#8221; Lagrangian constants<\/p><\/div>\n","protected":false},"excerpt":{"rendered":"<p>Probabilistic Graphical Models are generative models of complex systems. They rely on conditional independence assumptions between variables to learn sparse representations which can be visualized in a form of a graph. Such models are used for domain exploration and structure discovery in poorly understood domains. This work introduces a novel technique to perform sparse graph [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"footnotes":""},"msr-content-type":[3],"msr-research-highlight":[],"research-area":[13561,13556],"msr-publication-type":[193716],"msr-product-type":[],"msr-focus-area":[],"msr-platform":[],"msr-download-source":[],"msr-locale":[268875],"msr-field-of-study":[246694,246658,255316,246685],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-899655","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-algorithms","msr-research-area-artificial-intelligence","msr-locale-en_us","msr-field-of-study-artificial-intelligence","msr-field-of-study-deep-learning","msr-field-of-study-graph-neural-networks","msr-field-of-study-machine-learning"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2022-12-15","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"file","viewUrl":"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/uGLAD_model.pdf","id":"899667","title":"uglad_model","label_id":"243132","label":0}],"msr_related_uploader":"","msr_attachments":[{"id":1080186,"url":"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prodnew\/2024\/08\/uglad_architecture.pdf"},{"id":899667,"url":"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2022\/11\/uGLAD_model.pdf"}],"msr-author-ordering":[{"type":"user_nicename","value":"Harsh Shrivastava","user_id":41299,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Harsh Shrivastava"},{"type":"user_nicename","value":"Urszula Chajewska","user_id":38853,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Urszula Chajewska"},{"type":"user_nicename","value":"Robin Abraham","user_id":40819,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Robin Abraham"},{"type":"text","value":"Xinshi Chen","user_id":0,"rest_url":false}],"msr_impact_theme":[],"msr_research_lab":[199565],"msr_event":[],"msr_group":[],"msr_project":[771556],"publication":[],"video":[],"download":[],"msr_publication_type":"inproceedings","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/899655"}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":11,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/899655\/revisions"}],"predecessor-version":[{"id":1080261,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/899655\/revisions\/1080261"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=899655"}],"wp:term":[{"taxonomy":"msr-content-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-content-type?post=899655"},{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=899655"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=899655"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=899655"},{"taxonomy":"msr-product-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-product-type?post=899655"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=899655"},{"taxonomy":"msr-platform","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-platform?post=899655"},{"taxonomy":"msr-download-source","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-download-source?post=899655"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=899655"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=899655"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=899655"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=899655"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=899655"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=899655"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}