{"id":47227,"date":"2021-08-04T17:37:03","date_gmt":"2021-08-04T14:37:03","guid":{"rendered":"https:\/\/forklog.com\/en\/?p=47227"},"modified":"2025-09-01T20:14:46","modified_gmt":"2025-09-01T17:14:46","slug":"deepmind-develops-a-universal-architecture-for-building-artificial-intelligence","status":"publish","type":"post","link":"https:\/\/u1f987.com\/en\/deepmind-develops-a-universal-architecture-for-building-artificial-intelligence\/","title":{"rendered":"DeepMind develops a universal architecture for building artificial intelligence"},"content":{"rendered":"<p>Researchers at DeepMind have built the universal Perceiver IO architecture to handle all types of input and output data.<\/p>\n<p><!--more--><\/p>\n<blockquote class=\"twitter-tweet\">\n<p lang=\"en\" dir=\"ltr\">To tackle all the challenges we meet while solving intelligence, we need tools that are as adaptable as possible. Announcing the paper &#038; code for Perceiver IO, an architecture that handles a wide range of data and tasks, all while scaling gracefully: <a href=\"https:\/\/t.co\/9UU9b6276q\">https:\/\/t.co\/9UU9b6276q<\/a> 1\/4 <a href=\"https:\/\/t.co\/PO9JSHt7yL\">pic.twitter.com\/PO9JSHt7yL<\/a><\/p>\n<p>\u2014 DeepMind (@DeepMind) <a href=\"https:\/\/twitter.com\/DeepMind\/status\/1422590899343826944?ref_src=twsrc%5Etfw\">August 3, 2021<\/a><\/p><\/blockquote>\n<p> <script async=\"\" src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p>At its core is the original Perceiver model introduced in June 2021. It handles images, audio, video and their combinations, but is limited to tasks with simple outputs, such as classification.<\/p>\n<p>To address this limitation, researchers created a more general version of the architecture \u2014 Perceiver IO. It can produce a wide range of outputs from diverse input data, making it applicable to fields such as natural language processing, computer vision and multimodal understanding.<\/p>\n<figure class=\"wp-block-gallery columns-2 is-cropped wp-block-gallery-1 is-layout-flex wp-block-gallery-is-layout-flex\">\n<ul class=\"blocks-gallery-grid\">\n<li class=\"blocks-gallery-item\">\n<figure><img loading=\"lazy\" decoding=\"async\" width=\"480\" height=\"539\" src=\"https:\/\/u1f987.com\/wp-content\/uploads\/perceiver-io-deepmind.gif\" alt=\"DeepMind \u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0430\u043b\u0430 \u0443\u043d\u0438\u0432\u0435\u0440\u0441\u0430\u043b\u044c\u043d\u0443\u044e \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0443 \u0434\u043b\u044f \u0441\u043e\u0437\u0434\u0430\u043d\u0438\u044f \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442\u0430\" data-id=\"144718\" data-link=\"https:\/\/u1f987.com\/?attachment_id=144718\" class=\"wp-image-144718\"\/><\/figure>\n<\/li>\n<li class=\"blocks-gallery-item\">\n<figure><img loading=\"lazy\" decoding=\"async\" width=\"719\" height=\"719\" src=\"https:\/\/u1f987.com\/wp-content\/uploads\/perceiver-io-deepmind-2.png\" alt=\"DeepMind \u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0430\u043b\u0430 \u0443\u043d\u0438\u0432\u0435\u0440\u0441\u0430\u043b\u044c\u043d\u0443\u044e \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0443 \u0434\u043b\u044f \u0441\u043e\u0437\u0434\u0430\u043d\u0438\u044f \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442\u0430\" data-id=\"144719\" data-full-url=\"https:\/\/u1f987.com\/wp-content\/uploads\/perceiver-io-deepmind-2.png\" data-link=\"https:\/\/u1f987.com\/?attachment_id=144719\" class=\"wp-image-144719\" srcset=\"https:\/\/u1f987.com\/wp-content\/uploads\/perceiver-io-deepmind-2.png 719w, https:\/\/u1f987.com\/wp-content\/uploads\/perceiver-io-deepmind-2-300x300.png 300w, https:\/\/u1f987.com\/wp-content\/uploads\/perceiver-io-deepmind-2-150x150.png 150w\" sizes=\"auto, (max-width: 719px) 100vw, 719px\" \/><\/figure>\n<\/li>\n<\/ul><figcaption class=\"blocks-gallery-caption\">Results of optical flow or motion-tracking across all pixels of the image using Perceiver IO. The colour of each pixel shows the direction and speed of movement. Data.<a href=\"https:\/\/deepmind.com\/blog\/article\/building-architectures-that-can-handle-the-worlds-data\" target=\"_blank\" rel=\"noreferrer noopener nofollow\"> DeepMind<\/a>.<\/figcaption><\/figure>\n<p>Perceiver and Perceiver IO are built on transformer architectures that work well for input data containing thousands of elements. However, according to the researchers, images, audio and video can contain millions of such elements.<\/p>\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u201cWith the original Perceiver, we solved the main problem of a universal architecture: scaling transformers to very large input data without introducing domain-specific assumptions,\u201d the blog states.<\/p>\n<\/blockquote>\n<p>Researchers also believe that Perceiver IO could reach an unprecedented level of universality.<\/p>\n<p>They <a href=\"https:\/\/github.com\/deepmind\/deepmind-research\/tree\/master\/perceiver\" target=\"_blank\" rel=\"noreferrer noopener nofollow\">published<\/a> the architecture\u2019s source code on GitHub and hope it will help researchers and practitioners develop applications without having to spend resources on bespoke solutions built with specialized systems.<\/p>\n<p>Recall that in late July <a href=\"https:\/\/u1f987.com\/en\/news\/deepmind-builds-universal-ai-agents-in-an-open-ended-game\">DeepMind introduced the extensive gaming environment XLand for training<\/a> universal artificial intelligence agents.<\/p>\n<p>In July, the AI lab team <a href=\"https:\/\/u1f987.com\/en\/news\/deepmind-ai-maps-20000-human-protein-structures\">collected and published the most complete database of human protein structures<\/a>, created by the neural network AlphaFold.<\/p>\n<p>In June, scientists from DeepMind stated that <a href=\"https:\/\/u1f987.com\/en\/news\/deepmind-reinforcement-learning-suffices-to-create-general-ai\">reinforcement learning is sufficient to achieve artificial general intelligence<\/a>.<\/p>\n<p>Subscribe to ForkLog news on Telegram:<a href=\"https:\/\/t.me\/forklogAI\" target=\"_blank\" rel=\"noreferrer noopener nofollow\"> ForkLog AI<\/a> \u2014 all the news from the world of AI!<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Researchers at DeepMind have created the universal Perceiver IO architecture to process all types of input and output data.<\/p>\n","protected":false},"author":1,"featured_media":47228,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"select":"1","news_style_id":"1","cryptorium_level":"","_short_excerpt_text":"","creation_source":"","_metatest_mainpost_news_update":false,"footnotes":""},"categories":[3],"tags":[438,1474],"class_list":["post-47227","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-and-analysis","tag-artificial-intelligence","tag-deepmind"],"aioseo_notices":[],"amp_enabled":true,"views":"54","promo_type":"1","layout_type":"1","short_excerpt":"","is_update":"","_links":{"self":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/47227","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/comments?post=47227"}],"version-history":[{"count":1,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/47227\/revisions"}],"predecessor-version":[{"id":47229,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/posts\/47227\/revisions\/47229"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/media\/47228"}],"wp:attachment":[{"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/media?parent=47227"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/categories?post=47227"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/u1f987.com\/en\/wp-json\/wp\/v2\/tags?post=47227"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}