{"id":133780,"date":"2021-12-14T15:48:40","date_gmt":"2021-12-14T14:48:40","guid":{"rendered":"https:\/\/multi.datascientest.com\/?p=133780"},"modified":"2023-10-30T17:30:18","modified_gmt":"2023-10-30T16:30:18","slug":"recurrent-neural-network-rnn-de-que-se-trata","status":"publish","type":"post","link":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata","title":{"rendered":"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"133780\" class=\"elementor elementor-133780\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-9c9665d elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"9c9665d\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-3ea0302\" data-id=\"3ea0302\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-0f7fbc7 elementor-widget elementor-widget-text-editor\" data-id=\"0f7fbc7\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-text-editor.elementor-drop-cap-view-stacked .elementor-drop-cap{background-color:#69727d;color:#fff}.elementor-widget-text-editor.elementor-drop-cap-view-framed .elementor-drop-cap{color:#69727d;border:3px solid;background-color:transparent}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap{margin-top:8px}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap-letter{width:1em;height:1em}.elementor-widget-text-editor .elementor-drop-cap{float:left;text-align:center;line-height:1;font-size:50px}.elementor-widget-text-editor .elementor-drop-cap-letter{display:inline-block}<\/style>\t\t\t\t<h2><span style=\"color: #000000;\"><b data-stringify-type=\"bold\">\u00bfHas echado de menos a Daniel? \u00a1Nosotros tambi\u00e9n! Hoy, nuestro experto en Data Science que acompa\u00f1a a los alumnos a lo largo de su formaci\u00f3n responde a esta pregunta: \u00bfC\u00f3mo funcionan las redes neuronales recurrentes?<\/b><\/span><\/h2>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-bcd131f elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"bcd131f\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-4fc23c8\" data-id=\"4fc23c8\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-1232061 elementor-widget elementor-widget-text-editor\" data-id=\"1232061\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>Si eres un adepto a nuestro blog, ya sabr\u00e1s qu\u00e9 es una red neuronal (<b>si no es as\u00ed, no dudes en leer este art\u00edculo antes<\/b>), pero \u00bfqu\u00e9 aporta el adjetivo recurrente a este modelo? En este art\u00edculo vamos a ver c\u00f3mo se han convertido las redes neuronales recurrentes, llamadas RNN, en un modelo cl\u00e1sico dentro del Deep Learning.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-98f1406 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"98f1406\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-8ab9cc4\" data-id=\"8ab9cc4\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-24c7ecf elementor-widget elementor-widget-heading\" data-id=\"24c7ecf\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-heading-title{padding:0;margin:0;line-height:1}.elementor-widget-heading .elementor-heading-title[class*=elementor-size-]>a{color:inherit;font-size:inherit;line-height:inherit}.elementor-widget-heading .elementor-heading-title.elementor-size-small{font-size:15px}.elementor-widget-heading .elementor-heading-title.elementor-size-medium{font-size:19px}.elementor-widget-heading .elementor-heading-title.elementor-size-large{font-size:29px}.elementor-widget-heading .elementor-heading-title.elementor-size-xl{font-size:39px}.elementor-widget-heading .elementor-heading-title.elementor-size-xxl{font-size:59px}<\/style><h3 class=\"elementor-heading-title elementor-size-default\">Puesta en situaci\u00f3n<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-06df8e0 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"06df8e0\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-f1a4352\" data-id=\"f1a4352\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-3e156c9 elementor-widget elementor-widget-text-editor\" data-id=\"3e156c9\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\tAntes de explicitar una RNN, centr\u00e9monos en una pelota. Efectivamente, en Machine Learning suele ser frecuente querer predecir la trayectoria de un objeto m\u00f3vil. Como nos muestra la figura 1, desde el punto de partida, la pelota puede ir en cualquier direcci\u00f3n. \u00bfC\u00f3mo podemos saber cu\u00e1l es el pr\u00f3ximo movimiento de la pelota?\n\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-4444f3e elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"4444f3e\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-01374b5\" data-id=\"01374b5\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-7fd1427 elementor-widget elementor-widget-image\" data-id=\"7fd1427\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-image{text-align:center}.elementor-widget-image a{display:inline-block}.elementor-widget-image a img[src$=\".svg\"]{width:48px}.elementor-widget-image img{vertical-align:middle;display:inline-block}<\/style>\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"656\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-1024x840.png\" class=\"attachment-large size-large wp-image-80700\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-1024x840.png 1024w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-300x246.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-768x630.png 768w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1.png 1349w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d2d3c86 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d2d3c86\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-d413642\" data-id=\"d413642\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-58d4675 elementor-widget elementor-widget-text-editor\" data-id=\"58d4675\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>Sin embargo, si tomamos la figura 2, es evidente decir que la pelota va a seguir yendo hacia la derecha, gracias a las trayectorias anteriores, que traducen un movimiento hacia la derecha.<\/p><p><span style=\"font-weight: 400;\">Hasta aqu\u00ed todo parece l\u00f3gico.<\/span><\/p><blockquote><p><i><span style=\"font-weight: 400;\">Al contrario que en la figura 1, tenemos m\u00e1s datos de entrenamiento adicionales, de ah\u00ed que seamos capaces de decidir el desplazamiento de la pelota.<\/span><\/i><\/p><\/blockquote><p><span style=\"font-weight: 400;\">De este modo, basta con darle a nuestra red neuronal los antiguos desplazamientos de la pelota y se acab\u00f3 el estudio. Sin embargo, \u00bfc\u00f3mo podemos elegir el n\u00famero de neuronas de entrada? Una trayectoria puede ser compartida de la manera que queramos. Ya sea 10 o 100.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-e3e487f elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"e3e487f\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-a5b8926\" data-id=\"a5b8926\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-cb99836 elementor-widget elementor-widget-image\" data-id=\"cb99836\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"243\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-1024x311.png\" class=\"attachment-large size-large wp-image-80699\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-1024x311.png 1024w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-300x91.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-768x234.png 768w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-1536x467.png 1536w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-2048x623.png 2048w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-6837ace elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6837ace\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-c93e34e\" data-id=\"c93e34e\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-d332dfb elementor-widget elementor-widget-text-editor\" data-id=\"d332dfb\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">La pelota azul oscuro representa la posici\u00f3n actual de la pelota y las pelotas azul claro representan las antiguas trayectorias de nuestra pelota. De este modo adivinamos que la pelota se dirige hacia la derecha.<\/span><\/p><p><span style=\"font-weight: 400;\">Ocultemos ese detalle por ahora, y fijemos el tama\u00f1o de nuestras muestras de entrada. Observemos la figura 3, \u00bfhacia d\u00f3nde se va a dirigir la pelota?\u00a0<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-a69c6c4 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"a69c6c4\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-16b4429\" data-id=\"16b4429\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-5da4560 elementor-widget elementor-widget-image\" data-id=\"5da4560\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"484\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-2-1024x619.png\" class=\"attachment-large size-large wp-image-80698\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-2-1024x619.png 1024w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-2-300x181.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-2-768x464.png 768w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere_Plan-de-travail-1-copie-2.png 1349w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-aa3c4ca elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"aa3c4ca\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-30dc139\" data-id=\"30dc139\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-8b80df5 elementor-widget elementor-widget-text-editor\" data-id=\"8b80df5\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Es cierto que el movimiento de la pelota es m\u00e1s complejo que en la figura 1, pero seguimos pudiendo decir que el siguiente desplazamiento de la pelota ser\u00e1 hacia arriba. \u00bfHar\u00e1 el modelo la misma predicci\u00f3n? Por desgracia no, ya que la red neuronal no piensa como nosotros. El modelo, al contrario que nosotros, no toma en cuenta <\/span><b>el v\u00ednculo entre las entradas<\/b><span style=\"font-weight: 400;\">. Las entradas no son independientes las unas de las otras, por lo que debemos conservar ese v\u00ednculo entre ellas cuando entrenamos a nuestra red neuronal. <\/span><\/p><p>Tenemos que paliar 2 problemas :<\/p><ul><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">El tama\u00f1o de nuestras muestras de entrada <\/span><b>que no es fijo<\/b><span style=\"font-weight: 400;\">.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Los datos de entrada <\/span><b>no est\u00e1n vinculados entre ellos<\/b><span style=\"font-weight: 400;\">.<\/span><\/li><\/ul><p><b>Transici\u00f3n hacia la RNN<\/b><\/p><p><span style=\"font-weight: 400;\">Volvamos a nuestro estudio de trayectoria de pelota. Vamos a considerar un plan, de ese modo la trayectoria de la pelota tendr\u00e1 dos coordenadas, que llamaremos x^1,x^2 y queremos predecir la pr\u00f3xima ubicaci\u00f3n de la pelota, la predicci\u00f3n de las futuras coordenadas ser\u00e1 \u0177^1,\u0177^2. Vamos a representarlo con una red neuronal tradicional.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-0f9d6db elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"0f9d6db\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-2e55afa\" data-id=\"2e55afa\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-0e642dc elementor-widget elementor-widget-image\" data-id=\"0e642dc\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"365\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-04.png\" class=\"attachment-large size-large wp-image-80697\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-04.png 994w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-04-300x137.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-04-768x350.png 768w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d6bc7a9 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d6bc7a9\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-7cf4c3d\" data-id=\"7cf4c3d\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-16076de elementor-widget elementor-widget-text-editor\" data-id=\"16076de\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">De manera sint\u00e9tica, si planteamos (x^1,x^2)=x_t y (\u0177^1,\u0177^2)=\u0177_t, podemos hacer el siguiente esquema de abajo:<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-2d5fd1e elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2d5fd1e\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-5713026\" data-id=\"5713026\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-cff2b93 elementor-widget elementor-widget-image\" data-id=\"cff2b93\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"389\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-05-1.png\" class=\"attachment-large size-large wp-image-80692\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-05-1.png 984w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-05-1-300x146.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-05-1-768x374.png 768w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-976c237 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"976c237\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-078d0c4\" data-id=\"078d0c4\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-d3b5d80 elementor-widget elementor-widget-text-editor\" data-id=\"d3b5d80\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Aqu\u00ed el \u00edndice <\/span><i><span style=\"font-weight: 400;\">t<\/span><\/i><span style=\"font-weight: 400;\"> indica las coordenadas de la pelota en el momento <\/span><i><span style=\"font-weight: 400;\">t<\/span><\/i><span style=\"font-weight: 400;\">, representando la red neuronal con una funci\u00f3n <\/span><i><span style=\"font-weight: 400;\">f<\/span><\/i><span style=\"font-weight: 400;\">, nos queda :<\/span><\/p><p><span style=\"font-weight: 400;\">f(x_t)=\u0177_t (retomar el estilo matem\u00e1tico de la f\u00f3rmula RNN (abajo en ingl\u00e9s))<\/span><\/p><p><span style=\"font-weight: 400;\">Como hemos indicado antes, estudiamos el movimiento de la pelota de manera local. Queremos tener en cuenta varios instantes del movimiento de la pelota. Si no lo conseguimos con una red neuronal tradicional, con las RNN, todo cambia, porque se introduce el concepto de <\/span><b>recurrencia<\/b><span style=\"font-weight: 400;\">. Efectivamente, observemos la figura de abajo. A\u00f1adimos <\/span><b>una entrada<\/b> <i><span style=\"font-weight: 400;\">h_t<\/span><\/i><span style=\"font-weight: 400;\">, llamada <\/span><b>estado escondido<\/b><span style=\"font-weight: 400;\"> (hidden state). Ese estado escondido encarna <\/span><i><span style=\"font-weight: 400;\">\u0177_t<\/span><\/i><span style=\"font-weight: 400;\"> y se da como argumento para la pr\u00f3xima predicci\u00f3n adem\u00e1s de la entrada <\/span><i><span style=\"font-weight: 400;\">x_t<\/span><\/i><span style=\"font-weight: 400;\">. Efectivamente hemos considerado un conjunto y por tanto, vinculamos las salidas y las entradas sin l\u00edmite para nuestras muestras de entrada.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-44e2d88 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"44e2d88\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-089df2c\" data-id=\"089df2c\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-4d158c6 elementor-widget elementor-widget-image\" data-id=\"4d158c6\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"620\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-06-1024x793.png\" class=\"attachment-large size-large wp-image-80696\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-06-1024x793.png 1024w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-06-300x232.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-06-768x595.png 768w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-06-1536x1190.png 1536w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-06.png 1695w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-c0360f4 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"c0360f4\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-5fa2da7\" data-id=\"5fa2da7\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-d34cff3 elementor-widget elementor-widget-text-editor\" data-id=\"d34cff3\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Si tratamos de simbolizarlo con una f\u00f3rmula, obtenemos la siguiente ecuaci\u00f3n :\u00a0<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-0174b70 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"0174b70\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-5ce3e9d\" data-id=\"5ce3e9d\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-cc5897e elementor-widget elementor-widget-image\" data-id=\"cc5897e\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"760\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-08.png\" class=\"attachment-large size-large wp-image-80693\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-08.png 884w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-08-300x285.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-08-768x730.png 768w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-c006ea3 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"c006ea3\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-28c6395\" data-id=\"28c6395\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-f3019eb elementor-widget elementor-widget-text-editor\" data-id=\"f3019eb\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Aqu\u00ed se percibe bien el concepto de recurrencia. Para predecir el pr\u00f3ximo t\u00e9rmino, necesitamos informaci\u00f3n previa proporcionada por <\/span><i><span style=\"font-weight: 400;\">h_{t-1}<\/span><\/i><span style=\"font-weight: 400;\">.\u00a0<\/span><\/p><p><span style=\"font-weight: 400;\">La recurrencia es a\u00fan m\u00e1s flagrante en este esquema recapitulativo.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-a2d6dc1 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"a2d6dc1\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-c80d2d9\" data-id=\"c80d2d9\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a9a3a3b elementor-widget elementor-widget-heading\" data-id=\"a9a3a3b\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">\nQuelques applications des RNN \n<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-dd7a3b6 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"dd7a3b6\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-fc8d818\" data-id=\"fc8d818\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-28927fc elementor-widget elementor-widget-image\" data-id=\"28927fc\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"335\" src=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-07-1024x429.png\" class=\"attachment-large size-large wp-image-80695\" alt=\"\" srcset=\"https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-07-1024x429.png 1024w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-07-300x126.png 300w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-07-768x322.png 768w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-07-1536x644.png 1536w, https:\/\/datascientest.com\/es\/files\/2021\/07\/Sans-titre-1-Recupere-07-2048x859.png 2048w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-3a8bde9 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"3a8bde9\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-a1fa9fd\" data-id=\"a1fa9fd\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-704c2b7 elementor-widget elementor-widget-text-editor\" data-id=\"704c2b7\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<ul><li style=\"font-weight: 400;\" aria-level=\"1\"><i><span style=\"font-weight: 400;\">one to many<\/span><\/i><span style=\"font-weight: 400;\">: la RNN recibe <\/span><b>una \u00fanica entrada<\/b><span style=\"font-weight: 400;\"> y devuelve <\/span><b>varias salidas<\/b><span style=\"font-weight: 400;\">,<\/span><\/li><\/ul><p><span style=\"font-weight: 400;\">el ejemplo cl\u00e1sico de este procedimiento es la leyenda de la imagen.<\/span><\/p><ul><li style=\"font-weight: 400;\" aria-level=\"1\"><i><span style=\"font-weight: 400;\">many to <\/span><\/i><span style=\"font-weight: 400;\">one: tenemos varias entradas y solo hay una salida. Una ilustraci\u00f3n de este modo es el <\/span><b>an\u00e1lisis de sentimiento en textos<\/b><span style=\"font-weight: 400;\">. Esto permite identificar un sentimiento a partir de un grupo de palabras, determinar la palabra que falta para terminar la frase recibida en entrada. Para aprender m\u00e1s sobre el an\u00e1lisis de sentimientos, consulta <\/span><span style=\"font-weight: 400;\">este art\u00edculo<\/span><span style=\"font-weight: 400;\">.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><i><span style=\"font-weight: 400;\">many to many<\/span><\/i><span style=\"font-weight: 400;\">: por ultimo, podemos tomar varias entradas y obtener varias salidas. No tenemos por qu\u00e9 tener el mismo n\u00famero de neuronas de entrada y de salida. Aqu\u00ed podemos citar, la traducci\u00f3n de textos, pero tambi\u00e9n ser ambiciosos y prever el fin de una obra musical a partir de su comienzo.<\/span><\/li><\/ul><p><span style=\"font-weight: 400;\">Perfecto, hemos visto el funcionamiento de una RNN y sus m\u00faltiples aplicaciones, pero \u00bfes perfecto? Por desgracia, presenta un gran inconveniente, llamado <\/span><b>memoria a corto plazo<\/b><span style=\"font-weight: 400;\">, de la que veremos un ejemplo en TALN (Tratamiento autom\u00e1tico del lenguaje natural).<\/span><\/p><ul><li aria-level=\"1\"><p><b>\u00bfTiene la RNN memoria de pez?<\/b><\/p><p><span style=\"font-weight: 400;\">Retomemos el caso de completa una frase.\u00a0<\/span><\/p><p><i><span style=\"font-weight: 400;\">Me gusta el sushi, me voy a comer a<\/span><\/i><span style=\"font-weight: 400;\">\u2026<\/span><\/p><\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-ccad336 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"ccad336\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-e05f48c\" data-id=\"e05f48c\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-11534f5 elementor-widget elementor-widget-text-editor\" data-id=\"11534f5\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">La RNN no es capaz de retener la palabra sushi para predecir Jap\u00f3n, ya que no ha retenido la palabra sushi. Para llegar a determinar Jap\u00f3n, es necesario que la RNN posea una memoria mejor. Para ello hay que conseguir que las neuronas sean m\u00e1s complejas. En particular, vamos a ver el caso de la <\/span><b>LSTM<\/b><span style=\"font-weight: 400;\"> (<\/span><i><span style=\"font-weight: 400;\">Long Short Term Memory<\/span><\/i><span style=\"font-weight: 400;\">). Adem\u00e1s del estado escondido convencional <\/span><i><span style=\"font-weight: 400;\">h_t<\/span><\/i><span style=\"font-weight: 400;\">, vamos a a\u00f1adir un segundo estado <\/span><i><span style=\"font-weight: 400;\">c_t<\/span><\/i><span style=\"font-weight: 400;\">. Aqu\u00ed, <\/span><i><span style=\"font-weight: 400;\">h_t<\/span><\/i><span style=\"font-weight: 400;\"> representa la <\/span><b>memoria corta<\/b><span style=\"font-weight: 400;\"> neuronal y <\/span><i><span style=\"font-weight: 400;\">c_t<\/span><\/i><span style=\"font-weight: 400;\"> la <\/span><b>memoria larga<\/b><span style=\"font-weight: 400;\">.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-444c262 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"444c262\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-14c5c38\" data-id=\"14c5c38\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-0fbf6fb elementor-widget elementor-widget-text-editor\" data-id=\"0fbf6fb\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">No vamos a entrar en las consideraciones t\u00e9cnicas de este esquema aterrador. Lo esencial es observar que tenemos una c\u00e9lula mucho m\u00e1s compleja, que nos permite solucionar el problema de memoria. Con la LSTM, vamos a pasar las h_t y c_t por <\/span><b>puertas<\/b><span style=\"font-weight: 400;\">, 4 en total.<\/span><\/p><ul><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">La primera puerta permite eliminar la informaci\u00f3n in\u00fatil, es la <\/span><b>forget gate<\/b><span style=\"font-weight: 400;\">.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">La segunda puerta, almacena la informaci\u00f3n nueva, la <\/span><b>store gate<\/b><span style=\"font-weight: 400;\">.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">La tercera puerta actualiza la informaci\u00f3n que vamos a dar a la RNN con el resultado de la forget gate y la store gate, es la <\/span><b>update gate<\/b><span style=\"font-weight: 400;\">.<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Y la \u00faltima puerta (<\/span><b>output gate<\/b><span style=\"font-weight: 400;\">), nos da <\/span><i><span style=\"font-weight: 400;\">y_t <\/span><\/i><span style=\"font-weight: 400;\">y <\/span><i><span style=\"font-weight: 400;\">h_t<\/span><\/i><span style=\"font-weight: 400;\">.<\/span><\/li><\/ul><p><span style=\"font-weight: 400;\">Este largo proceso permite controlar la informaci\u00f3n que conservamos y transmitimos a trav\u00e9s del tiempo.<\/span><\/p><p><span style=\"font-weight: 400;\">La RNN llega a saber qu\u00e9 hay que guardar y qu\u00e9 hay que olvidar, gracias a su aprendizaje. La <\/span><b>LSTM<\/b><span style=\"font-weight: 400;\"> (<\/span><i><span style=\"font-weight: 400;\">Long Short Term Memory<\/span><\/i><span style=\"font-weight: 400;\">) no es \u00fanica, tambi\u00e9n podemos usar <\/span><b>GRU<\/b><span style=\"font-weight: 400;\"> (<\/span><i><span style=\"font-weight: 400;\">Gated Recurrent Unit<\/span><\/i><span style=\"font-weight: 400;\">), solo cambia la arquitectura de la c\u00e9lula.<\/span><\/p><p><span style=\"font-weight: 400;\">Ahora resumamos lo que hemos visto. Las RNN son un tipo particular de redes neuronales que permiten tratar datos que no son <\/span><b>independientes<\/b><span style=\"font-weight: 400;\"> y que no tienen un <\/span><b>tama\u00f1o fijo<\/b><span style=\"font-weight: 400;\">. No obstante. las RNN est\u00e1ndar est\u00e1n bastante limitadas con el problema de <\/span><b>memoria corta<\/b><span style=\"font-weight: 400;\">, pero lo podemos resolver utilizando c\u00e9lulas m\u00e1s complejas como las <\/span><b>LSTM<\/b><span style=\"font-weight: 400;\"> o <\/span><b>GRU<\/b><span style=\"font-weight: 400;\">.\u00a0<\/span><\/p><p><span style=\"font-weight: 400;\">Adem\u00e1s, podemos hacer un paralelismo con otro sistema de red neuronal: las <\/span><b>redes neuronales convolucionales<\/b><span style=\"font-weight: 400;\"> (CNN en ingl\u00e9s). Efectivamente las CNN son conocidas por compartir <\/span><b>la informaci\u00f3n espacial<\/b><span style=\"font-weight: 400;\"> mientras que las RNN comparten la <\/span><b>informaci\u00f3n temporal<\/b><span style=\"font-weight: 400;\">. Encontrar\u00e1s m\u00e1s informaci\u00f3n sobre las CNN en <\/span><a href=\"https:\/\/datascientest.com\/convolutional-neural-network\"><span style=\"font-weight: 400;\">este art\u00edculo<\/span><\/a><span style=\"font-weight: 400;\">. Y por \u00faltimo, si quieres poner en pr\u00e1ctica las RNN, no dudes en apuntarte a nuestro curso de Data Scientist.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-60f3416 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"60f3416\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-5472242\" data-id=\"5472242\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-567be99 elementor-align-center elementor-widget elementor-widget-button\" data-id=\"567be99\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-md elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formations-data\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">Descubrir el curso<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>\u00bfHas echado de menos a Daniel? \u00a1Nosotros tambi\u00e9n! Hoy, nuestro experto en Data Science que acompa\u00f1a a los alumnos a lo largo de su formaci\u00f3n responde a esta pregunta: \u00bfC\u00f3mo funcionan las redes neuronales recurrentes? Si eres un adepto a nuestro blog, ya sabr\u00e1s qu\u00e9 es una red neuronal (si no es as\u00ed, no dudes [&hellip;]<\/p>\n","protected":false},"author":45,"featured_media":80135,"comment_status":"open","ping_status":"open","sticky":false,"template":"elementor_theme","format":"standard","meta":{"content-type":"","inline_featured_image":false},"categories":[2427],"tags":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v21.7 (Yoast SEO v21.7) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?<\/title>\n<meta name=\"description\" content=\"Si eres seguidor de nuestro blog, ya sabes lo que es una red neuronal (si no, no dudes en leer este art\u00edculo primero) pero \u00bfqu\u00e9 aporta el adjetivo recurrente a este modelo? En este art\u00edculo veremos c\u00f3mo las redes neuronales recurrentes, llamadas RNN, se han convertido en un modelo cl\u00e1sico en deep learning.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata\" \/>\n<meta property=\"og:locale\" content=\"es_ES\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?\" \/>\n<meta property=\"og:description\" content=\"\u00bfHas echado de menos a Daniel? \u00a1Nosotros tambi\u00e9n! Hoy, nuestro experto en Data Science que acompa\u00f1a a los alumnos a lo largo de su formaci\u00f3n responde a\" \/>\n<meta property=\"og:url\" content=\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata\" \/>\n<meta property=\"og:site_name\" content=\"Formaci\u00f3n en ciencia de datos | DataScientest.com\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/DataScientestEN\" \/>\n<meta property=\"article:published_time\" content=\"2021-12-14T14:48:40+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-10-30T16:30:18+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/datascientest.com\/es\/wp-content\/blogs.dir\/7\/files\/2021\/06\/illu_reseaux_neurones-25.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1024\" \/>\n\t<meta property=\"og:image:height\" content=\"562\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Daniel\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@DataScientest__\" \/>\n<meta name=\"twitter:site\" content=\"@DataScientest__\" \/>\n<meta name=\"twitter:label1\" content=\"Escrito por\" \/>\n\t<meta name=\"twitter:data1\" content=\"Daniel\" \/>\n\t<meta name=\"twitter:label2\" content=\"Tiempo de lectura\" \/>\n\t<meta name=\"twitter:data2\" content=\"8 minutos\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata#article\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata\"},\"author\":{\"name\":\"Daniel\",\"@id\":\"https:\/\/datascientest.com\/es\/#\/schema\/person\/5a86e30c5090234fb43dd8baa83c44b0\"},\"headline\":\"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?\",\"datePublished\":\"2021-12-14T14:48:40+00:00\",\"dateModified\":\"2023-10-30T16:30:18+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata\"},\"wordCount\":1364,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\/\/datascientest.com\/es\/#organization\"},\"articleSection\":[\"Data Science\"],\"inLanguage\":\"es\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata#respond\"]}],\"copyrightYear\":\"2021\",\"copyrightHolder\":{\"@id\":\"https:\/\/datascientest.com\/#organization\"}},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata\",\"url\":\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata\",\"name\":\"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/es\/#website\"},\"datePublished\":\"2021-12-14T14:48:40+00:00\",\"dateModified\":\"2023-10-30T16:30:18+00:00\",\"description\":\"Si eres seguidor de nuestro blog, ya sabes lo que es una red neuronal (si no, no dudes en leer este art\u00edculo primero) pero \u00bfqu\u00e9 aporta el adjetivo recurrente a este modelo? En este art\u00edculo veremos c\u00f3mo las redes neuronales recurrentes, llamadas RNN, se han convertido en un modelo cl\u00e1sico en deep learning.\",\"inLanguage\":\"es\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata\"]}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/datascientest.com\/es\/#website\",\"url\":\"https:\/\/datascientest.com\/es\/\",\"name\":\"Formaci\u00f3n en ciencia de datos | DataScientest.com\",\"description\":\"L\u00edder en formaci\u00f3n en Ciencia de Datos\",\"publisher\":{\"@id\":\"https:\/\/datascientest.com\/es\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/datascientest.com\/es\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"es\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/datascientest.com\/es\/#organization\",\"name\":\"DataScientest\",\"url\":\"https:\/\/datascientest.com\/es\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"es\",\"@id\":\"https:\/\/datascientest.com\/es\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/datascientest.com\/es\/wp-content\/blogs.dir\/7\/files\/2024\/02\/DataScientest_icon_blue-2.png\",\"contentUrl\":\"https:\/\/datascientest.com\/es\/wp-content\/blogs.dir\/7\/files\/2024\/02\/DataScientest_icon_blue-2.png\",\"width\":301,\"height\":301,\"caption\":\"DataScientest\"},\"image\":{\"@id\":\"https:\/\/datascientest.com\/es\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.facebook.com\/DataScientestEN\",\"https:\/\/twitter.com\/DataScientest__\",\"https:\/\/www.linkedin.com\/company\/datascientest-spain\/\",\"https:\/\/www.tiktok.com\/@datascientest\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/datascientest.com\/es\/#\/schema\/person\/5a86e30c5090234fb43dd8baa83c44b0\",\"name\":\"Daniel\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"es\",\"@id\":\"https:\/\/datascientest.com\/es\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/837acad4f2dc47e85d81cce837261670?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/837acad4f2dc47e85d81cce837261670?s=96&d=mm&r=g\",\"caption\":\"Daniel\"},\"sameAs\":[\"https:\/\/www.linkedin.com\/school\/datascientest\/\",\"user\"],\"jobTitle\":\"ing\u00e9nieur logiciel\",\"worksFor\":\"DataScientest\",\"url\":\"https:\/\/datascientest.com\/es\/author\/team-redac\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?","description":"Si eres seguidor de nuestro blog, ya sabes lo que es una red neuronal (si no, no dudes en leer este art\u00edculo primero) pero \u00bfqu\u00e9 aporta el adjetivo recurrente a este modelo? En este art\u00edculo veremos c\u00f3mo las redes neuronales recurrentes, llamadas RNN, se han convertido en un modelo cl\u00e1sico en deep learning.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata","og_locale":"es_ES","og_type":"article","og_title":"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?","og_description":"\u00bfHas echado de menos a Daniel? \u00a1Nosotros tambi\u00e9n! Hoy, nuestro experto en Data Science que acompa\u00f1a a los alumnos a lo largo de su formaci\u00f3n responde a","og_url":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata","og_site_name":"Formaci\u00f3n en ciencia de datos | DataScientest.com","article_publisher":"https:\/\/www.facebook.com\/DataScientestEN","article_published_time":"2021-12-14T14:48:40+00:00","article_modified_time":"2023-10-30T16:30:18+00:00","og_image":[{"width":1024,"height":562,"url":"https:\/\/datascientest.com\/es\/wp-content\/blogs.dir\/7\/files\/2021\/06\/illu_reseaux_neurones-25.png","type":"image\/png"}],"author":"Daniel","twitter_card":"summary_large_image","twitter_creator":"@DataScientest__","twitter_site":"@DataScientest__","twitter_misc":{"Escrito por":"Daniel","Tiempo de lectura":"8 minutos"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata#article","isPartOf":{"@id":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata"},"author":{"name":"Daniel","@id":"https:\/\/datascientest.com\/es\/#\/schema\/person\/5a86e30c5090234fb43dd8baa83c44b0"},"headline":"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?","datePublished":"2021-12-14T14:48:40+00:00","dateModified":"2023-10-30T16:30:18+00:00","mainEntityOfPage":{"@id":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata"},"wordCount":1364,"commentCount":0,"publisher":{"@id":"https:\/\/datascientest.com\/es\/#organization"},"articleSection":["Data Science"],"inLanguage":"es","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata#respond"]}],"copyrightYear":"2021","copyrightHolder":{"@id":"https:\/\/datascientest.com\/#organization"}},{"@type":"WebPage","@id":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata","url":"https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata","name":"Recurrent Neural Network (RNN): \u00bfde qu\u00e9 se trata?","isPartOf":{"@id":"https:\/\/datascientest.com\/es\/#website"},"datePublished":"2021-12-14T14:48:40+00:00","dateModified":"2023-10-30T16:30:18+00:00","description":"Si eres seguidor de nuestro blog, ya sabes lo que es una red neuronal (si no, no dudes en leer este art\u00edculo primero) pero \u00bfqu\u00e9 aporta el adjetivo recurrente a este modelo? En este art\u00edculo veremos c\u00f3mo las redes neuronales recurrentes, llamadas RNN, se han convertido en un modelo cl\u00e1sico en deep learning.","inLanguage":"es","potentialAction":[{"@type":"ReadAction","target":["https:\/\/datascientest.com\/es\/recurrent-neural-network-rnn-de-que-se-trata"]}]},{"@type":"WebSite","@id":"https:\/\/datascientest.com\/es\/#website","url":"https:\/\/datascientest.com\/es\/","name":"Formaci\u00f3n en ciencia de datos | DataScientest.com","description":"L\u00edder en formaci\u00f3n en Ciencia de Datos","publisher":{"@id":"https:\/\/datascientest.com\/es\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/datascientest.com\/es\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"es"},{"@type":"Organization","@id":"https:\/\/datascientest.com\/es\/#organization","name":"DataScientest","url":"https:\/\/datascientest.com\/es\/","logo":{"@type":"ImageObject","inLanguage":"es","@id":"https:\/\/datascientest.com\/es\/#\/schema\/logo\/image\/","url":"https:\/\/datascientest.com\/es\/wp-content\/blogs.dir\/7\/files\/2024\/02\/DataScientest_icon_blue-2.png","contentUrl":"https:\/\/datascientest.com\/es\/wp-content\/blogs.dir\/7\/files\/2024\/02\/DataScientest_icon_blue-2.png","width":301,"height":301,"caption":"DataScientest"},"image":{"@id":"https:\/\/datascientest.com\/es\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/DataScientestEN","https:\/\/twitter.com\/DataScientest__","https:\/\/www.linkedin.com\/company\/datascientest-spain\/","https:\/\/www.tiktok.com\/@datascientest"]},{"@type":"Person","@id":"https:\/\/datascientest.com\/es\/#\/schema\/person\/5a86e30c5090234fb43dd8baa83c44b0","name":"Daniel","image":{"@type":"ImageObject","inLanguage":"es","@id":"https:\/\/datascientest.com\/es\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/837acad4f2dc47e85d81cce837261670?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/837acad4f2dc47e85d81cce837261670?s=96&d=mm&r=g","caption":"Daniel"},"sameAs":["https:\/\/www.linkedin.com\/school\/datascientest\/","user"],"jobTitle":"ing\u00e9nieur logiciel","worksFor":"DataScientest","url":"https:\/\/datascientest.com\/es\/author\/team-redac"}]}},"_links":{"self":[{"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/posts\/133780"}],"collection":[{"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/users\/45"}],"replies":[{"embeddable":true,"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/comments?post=133780"}],"version-history":[{"count":0,"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/posts\/133780\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/media\/80135"}],"wp:attachment":[{"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/media?parent=133780"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/categories?post=133780"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/datascientest.com\/es\/wp-json\/wp\/v2\/tags?post=133780"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}