{"id":77,"date":"2017-06-30T20:25:21","date_gmt":"2017-06-30T20:25:21","guid":{"rendered":"http:\/\/www.inf.udec.cl\/~guille\/?page_id=77"},"modified":"2025-01-25T14:32:56","modified_gmt":"2025-01-25T14:32:56","slug":"research","status":"publish","type":"page","link":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/research\/","title":{"rendered":"Research"},"content":{"rendered":"<div class=\"su-box su-box-style-default\" id=\"\" style=\"border-color:#b9c5cc;border-radius:3px;max-width:none\"><div class=\"su-box-title\" style=\"background-color:#ecf8ff;color:#505050;border-top-left-radius:1px;border-top-right-radius:1px\">Artificial Intelligences for Variable Astronomical Sources<\/div><div class=\"su-box-content su-u-clearfix su-u-trim\" style=\"border-bottom-left-radius:1px;border-bottom-right-radius:1px\">\n<p><span class=\"s1\">During the last couple of years, astronomy has become a\u00a0<\/span><span class=\"s1\">data-driven science. We have evolved from the Megabyte to the Gigabyte\u00a0<\/span><span class=\"s1\">and to the Terabyte regime in less than a decade. Thanks to the\u00a0<\/span><span class=\"s1\">special characteristics of the skies in Chile (low humidity, high\u00a0<\/span><span class=\"s1\">peaks and plains, low light pollution, and large number of clear\u00a0<\/span><span class=\"s1\">nights), most of the major observatories are being installed here.<\/span><span class=\"s1\"> Survey\u00a0<\/span><span class=\"s1\">telescopes, such as the\u00a0<\/span><span class=\"s1\">Large Synoptic Survey Telescope (<\/span><span class=\"s1\">LSST), scan a large area of the sky in order to\u00a0<\/span><span class=\"s1\">observe a significant amount of objects, including galaxies, stars, and asteroids,\u00a0<\/span><span class=\"s1\">among others.\u00a0<\/span><span class=\"s1\">We are applying prediction models over variable astronomical objects in order to characterize them and detect new unknown sources never seen before. We are applying different neural network architectures for classification and novelty detection of light curves and sequences of images.<\/span><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-131 aligncenter\" src=\"http:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-20-at-5.31.43-PM.png\" alt=\"\" width=\"511\" height=\"309\" srcset=\"https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-20-at-5.31.43-PM.png 888w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-20-at-5.31.43-PM-300x181.png 300w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-20-at-5.31.43-PM-768x464.png 768w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-20-at-5.31.43-PM-750x454.png 750w\" sizes=\"auto, (max-width: 511px) 100vw, 511px\" \/><\/p>\n<\/div><\/div>\n<p>&nbsp;<\/p>\n<div class=\"su-box su-box-style-default\" id=\"\" style=\"border-color:#b9c5cc;border-radius:3px;max-width:none\"><div class=\"su-box-title\" style=\"background-color:#ecf8ff;color:#505050;border-top-left-radius:1px;border-top-right-radius:1px\">A.I. to support diagnosis for patients with cancer<\/div><div class=\"su-box-content su-u-clearfix su-u-trim\" style=\"border-bottom-left-radius:1px;border-bottom-right-radius:1px\">\n<div title=\"Page 1\">Many machine learning models have been developed for patients with cancer, but integrating and validating them in clinical processes has proven to be hard. We are evaluating the applicability of state-of-the-art predictive models \u00a0in a real medical scenario, as well as developing new predictive models to aid radiologists and\u00a0oncologists.<\/div>\n<div title=\"Page 1\"><\/div>\n<div title=\"Page 1\"><\/div>\n<div title=\"Page 1\"><img loading=\"lazy\" decoding=\"async\" class=\" wp-image-337 aligncenter\" src=\"http:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screenshot-2025-01-25-at-11.30.21\u202fAM.png\" alt=\"\" width=\"499\" height=\"181\" srcset=\"https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screenshot-2025-01-25-at-11.30.21\u202fAM.png 744w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screenshot-2025-01-25-at-11.30.21\u202fAM-300x109.png 300w\" sizes=\"auto, (max-width: 499px) 100vw, 499px\" \/><\/div>\n<div title=\"Page 1\"><\/div>\n<div title=\"Page 1\"><\/div>\n<div title=\"Page 1\"><\/div>\n<div title=\"Page 1\"><\/div>\n<div class=\"page\" title=\"Page 1\">\u00a0<\/div><\/div><\/div>\n<div class=\"su-box su-box-style-default\" id=\"\" style=\"border-color:#b9c5cc;border-radius:3px;max-width:none\"><div class=\"su-box-title\" style=\"background-color:#ecf8ff;color:#505050;border-top-left-radius:1px;border-top-right-radius:1px\">Biased Data<\/div><div class=\"su-box-content su-u-clearfix su-u-trim\" style=\"border-bottom-left-radius:1px;border-bottom-right-radius:1px\">\n<div class=\"page\" title=\"Page 1\">\n<div class=\"layoutArea\">\n<div class=\"column\">\n<p>Supervised machine learning usually relies on labels created by human annotators. This labels, called the ground truth are assumed to be correct or contain random white noise errors. However, real data is never perfect, and the process of labeling can be systematically biased due to the quality of the observed data. For example, when labeling images, human labels can be biased in terms of the image resolution, making all humans to agree in an incorrect label. The variance on the labels can be very small and the estimated labels still be wrong.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\" wp-image-125 aligncenter\" src=\"http:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2018\/08\/Screen-Shot-2018-08-20-at-5.16.24-PM.png\" alt=\"\" width=\"362\" height=\"224\" srcset=\"https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2018\/08\/Screen-Shot-2018-08-20-at-5.16.24-PM.png 481w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2018\/08\/Screen-Shot-2018-08-20-at-5.16.24-PM-300x186.png 300w\" sizes=\"auto, (max-width: 362px) 100vw, 362px\" \/><\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div><\/div>\n<div class=\"su-box su-box-style-default\" id=\"\" style=\"border-color:#b9c5cc;border-radius:3px;max-width:none\"><div class=\"su-box-title\" style=\"background-color:#ecf8ff;color:#505050;border-top-left-radius:1px;border-top-right-radius:1px\">Deep domain adaptation<\/div><div class=\"su-box-content su-u-clearfix su-u-trim\" style=\"border-bottom-left-radius:1px;border-bottom-right-radius:1px\">\n<p class=\"p1\"><span class=\"s1\">Deep learning methods for classification are more accurate than\u00a0<\/span><span class=\"s1\">feature engineering approaches but require more labeled data to\u00a0<\/span><span class=\"s1\">achieve such performances. We are exploring new ways of training a\u00a0 deep learning architecture\u00a0<\/span><span class=\"s1\">on a\u00a0<\/span><em><span class=\"s3\">source<\/span><\/em><span class=\"s1\">\u00a0labeled dataset and adapt it to a differently\u00a0<\/span><span class=\"s1\">distributed <\/span><em><span class=\"s3\">target<\/span><\/em><span class=\"s1\">\u00a0unlabeled dataset. This problem is particularly relevant for astronomy and genomic data, as new datasets are constantly being created and their labeling process is expensive.<\/span><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-170 aligncenter\" src=\"http:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-21-at-8.45.24-PM-1024x427.png\" alt=\"\" width=\"612\" height=\"255\" srcset=\"https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-21-at-8.45.24-PM-1024x427.png 1024w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-21-at-8.45.24-PM-300x125.png 300w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-21-at-8.45.24-PM-768x320.png 768w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-21-at-8.45.24-PM-750x312.png 750w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-21-at-8.45.24-PM.png 1205w\" sizes=\"auto, (max-width: 612px) 100vw, 612px\" \/><\/p>\n<\/div><\/div>\n<p>&nbsp;<\/p>\n<div class=\"su-box su-box-style-default\" id=\"\" style=\"border-color:#b9c5cc;border-radius:3px;max-width:none\"><div class=\"su-box-title\" style=\"background-color:#ecf8ff;color:#505050;border-top-left-radius:1px;border-top-right-radius:1px\">Deep learning models for identifying cancer pathways<\/div><div class=\"su-box-content su-u-clearfix su-u-trim\" style=\"border-bottom-left-radius:1px;border-bottom-right-radius:1px\">\n<p>We are analysing single-cell genome sequencing (scRNAseq) data to identify specific pathways associated to different types of cancer. We found that the regulatory T cells reveal a unique genetic signature across them. Although we found a set of other cell types, our focus is on T cells because they modulate the immune system, controlling immune responses. We are currently developing\u00a0 deep learning models in order to integrate the different types of cells for different types of cancer through a semi-supervised learning approach.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-206 aligncenter\" src=\"http:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-22-at-10.07.03-PM.png\" alt=\"\" width=\"381\" height=\"413\" srcset=\"https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-22-at-10.07.03-PM.png 721w, https:\/\/www.inf.udec.cl\/~guille\/wp-content\/uploads\/2017\/06\/Screen-Shot-2018-08-22-at-10.07.03-PM-277x300.png 277w\" sizes=\"auto, (max-width: 381px) 100vw, 381px\" \/><\/p>\n<\/div><\/div>\n","protected":false},"excerpt":{"rendered":"<p>&nbsp; &nbsp;<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-77","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/pages\/77","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/comments?post=77"}],"version-history":[{"count":24,"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/pages\/77\/revisions"}],"predecessor-version":[{"id":339,"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/pages\/77\/revisions\/339"}],"wp:attachment":[{"href":"https:\/\/www.inf.udec.cl\/~guille\/index.php\/wp-json\/wp\/v2\/media?parent=77"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}