{"id":7713,"date":"2025-10-20T07:03:53","date_gmt":"2025-10-20T07:03:53","guid":{"rendered":"https:\/\/mailitics.com\/index.php\/2025\/10\/20\/transformers_time_series_and_the_myth_of\/"},"modified":"2025-10-20T07:03:53","modified_gmt":"2025-10-20T07:03:53","slug":"transformers_time_series_and_the_myth_of","status":"publish","type":"post","link":"https:\/\/mailitics.com\/index.php\/2025\/10\/20\/transformers_time_series_and_the_myth_of\/","title":{"rendered":"Transformers, Time Series, and the Myth of Permutation Invariance"},"content":{"rendered":"<p>    Transformers, Time Series, and the Myth of Permutation Invariance<br \/>\n \t<BR><br \/>\n<BR><\/BR><br \/>\n    <!-- no image --><br \/>\n \t<BR><br \/>\n<BR><\/BR><\/p>\n<div>\n<!-- SC_OFF --><\/p>\n<div class=\"md\">\n<p>There&#8217;s a common misconception in ML\/DL that <em>Transformers shouldn\u2019t be used for forecasting because attention is permutation-invariant.<\/em><\/p>\n<p>Latest evidence shows the opposite, such as Google&#8217;s latest model, where the experiments show the model performs just as well with or without positional embeddings.<\/p>\n<p>You can find an analysis on tis topic <a href=\"https:\/\/aihorizonforecast.substack.com\/p\/transformers-time-series-and-the\">here<\/a>.<\/p>\n<\/p><\/div>\n<p><!-- SC_ON -->   submitted by   <a href=\"https:\/\/www.reddit.com\/user\/nkafr\"> \/u\/nkafr <\/a> <br \/> <span><a href=\"https:\/\/www.reddit.com\/r\/datascience\/comments\/1oa6dn1\/transformers_time_series_and_the_myth_of\/\">[link]<\/a><\/span>   <span><a href=\"https:\/\/www.reddit.com\/r\/datascience\/comments\/1oa6dn1\/transformers_time_series_and_the_myth_of\/\">[comments]<\/a><\/span>\n<\/div>\n<p> \t<BR><br \/>\n <BR><\/BR><br \/>\n    \/u\/nkafr<br \/>\n \t<BR><br \/>\n<BR><\/BR><br \/>\n<a href=\"https:\/\/www.reddit.com\/r\/datascience\/comments\/1oa6dn1\/transformers_time_series_and_the_myth_of\/\">Go to original source<\/a><br \/>\n \t<BR><br \/>\n <BR><\/BR><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Transformers, Time Series, and the Myth of Permutation Invariance There&#8217;s a common misconception in ML\/DL that Transformers shouldn\u2019t be used for forecasting because attention is permutation-invariant. Latest evidence shows the opposite, such as Google&#8217;s latest model, where the experiments show the model performs just as well with or without positional embeddings. You can find an [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[62,99],"tags":[1993,15,648],"class_list":["post-7713","post","type-post","status-publish","format-standard","hentry","category-aimldsaimlds","category-datascience","tag-permutation","tag-time","tag-transformers"],"_links":{"self":[{"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/posts\/7713"}],"collection":[{"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/comments?post=7713"}],"version-history":[{"count":0,"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/posts\/7713\/revisions"}],"wp:attachment":[{"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/media?parent=7713"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/categories?post=7713"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/mailitics.com\/index.php\/wp-json\/wp\/v2\/tags?post=7713"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}