{"id":3029,"date":"2017-08-31T22:54:29","date_gmt":"2017-08-31T20:54:29","guid":{"rendered":"https:\/\/blog.mi.hdm-stuttgart.de\/?p=3029"},"modified":"2023-06-08T17:35:28","modified_gmt":"2023-06-08T15:35:28","slug":"wettersave-realizing-weather-forecasts-with-machine-learning","status":"publish","type":"post","link":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2017\/08\/31\/wettersave-realizing-weather-forecasts-with-machine-learning\/","title":{"rendered":"Wettersave \u2013 Realizing Weather Forecasts with Machine Learning"},"content":{"rendered":"<p><strong>&nbsp;&nbsp;&nbsp;&nbsp;Introduction<\/strong><\/p>\n<p>Since the internet boom a few years ago companies started to collect and save data in an almost aggressive way. But the huge amounts of data are actually useless if they are not used to gain new information with a higher value. I was always impressed by the way how easy statistical algorithms can be used to answer extremely complex questions if you add the component \u201cMachine Learning\u201d. Our goal was to create a web service that does exactly that thing: We realized Weather Forecasts using Machine Learning algorithms.<\/p>\n<p>The Application can be split in four parts:<\/p>\n<ul>\n<li>The website is the final user interface to start a query and to see the resulting prediction.<\/li>\n<li>The server connects all parts and hosts the website.<\/li>\n<li>The database stores all important data about the weather.<\/li>\n<li>IBM Watson is used to calculate the forecasts with the data of the database.<\/li>\n<\/ul>\n<p>In the following I will explain the structure more detailed and show how we developed the application.<br \/>\n<!--more--><br \/>\n<strong>Development<\/strong><\/p>\n<p><strong>The Database<\/strong><\/p>\n<p>First, we needed the right data about the weather to start with. The idea was to connect our application with the weather API hosted at IBM Bluemix. Unfortunately, it did not work out because first the data we found there was useless for our aimed predictions and second the service was too expensive. So, what we did is, we took the free weather data provided by the DWD (Deutscher Wetter Dienst) and saved it at our own database hosted at Bluemix. The data includes all important information from 12\/23\/2015 until 6\/20\/2017 so only predictions in this range are logically possible and comparable to the actual values.<\/p>\n<p><strong>Machine Learning with IBM Watson<\/strong><\/p>\n<p>We created the prediction models with the SPSS Modeler by IBM. These so-called streams can be uploaded directly to the Watson Machine Learning Service. To create a stream based on our data, we first had to connect the database we used to train the model with the SPSS Modeler. The next step was to filter the data and to leave out all information that was irrelevant for creating the prediction model, for example the ID of each record as it has no influence on the predicted values. The records contained in the database are adapted to the model that predicts the weather for a chosen day using the information of one day before. This is why each record contains all measurements of one day and the measurements of the following one. Then all inputs and the relating outputs or results that will be predicted are marked at the field operation \u201ctype\u201d. The measurements that were not used in this prediction model are marked with the role \u201cnone\u201d to exclude them from the following computation.<\/p>\n<p>On the following picture you can see the filter operation (left) and the type operation (right):<\/p>\n<p><strong><a href=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp.jpg\"><img loading=\"lazy\" decoding=\"async\" data-attachment-id=\"3031\" data-permalink=\"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2017\/08\/31\/wettersave-realizing-weather-forecasts-with-machine-learning\/filterundtyp\/\" data-orig-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp.jpg\" data-orig-size=\"1574,708\" data-comments-opened=\"1\" data-image-meta=\"{&quot;aperture&quot;:&quot;0&quot;,&quot;credit&quot;:&quot;joel&quot;,&quot;camera&quot;:&quot;&quot;,&quot;caption&quot;:&quot;&quot;,&quot;created_timestamp&quot;:&quot;1504207047&quot;,&quot;copyright&quot;:&quot;&quot;,&quot;focal_length&quot;:&quot;0&quot;,&quot;iso&quot;:&quot;0&quot;,&quot;shutter_speed&quot;:&quot;0&quot;,&quot;title&quot;:&quot;&quot;,&quot;orientation&quot;:&quot;0&quot;}\" data-image-title=\"filterUndTyp\" data-image-description=\"\" data-image-caption=\"\" data-large-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp-1024x461.jpg\" class=\"alignnone  wp-image-3031\" src=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp-300x135.jpg\" alt=\"\" width=\"433\" height=\"195\" srcset=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp-300x135.jpg 300w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp-768x345.jpg 768w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp-1024x461.jpg 1024w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/filterUndTyp.jpg 1574w\" sizes=\"auto, (max-width: 433px) 100vw, 433px\" \/><\/a><\/strong><\/p>\n<p>The Data was now ready for modelling and the auto numeric operation of the SPSS Modeler could calculate the models. The auto numeric operation calculates different kind of model types and chooses the one with the best results. For the prediction based on one previous day a neural network was defined to be the model with the best results.<\/p>\n<p><strong><a href=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/modellzusammenfassung.jpg\"><img loading=\"lazy\" decoding=\"async\" data-attachment-id=\"3032\" data-permalink=\"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2017\/08\/31\/wettersave-realizing-weather-forecasts-with-machine-learning\/modellzusammenfassung\/\" data-orig-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/modellzusammenfassung.jpg\" data-orig-size=\"483,381\" data-comments-opened=\"1\" data-image-meta=\"{&quot;aperture&quot;:&quot;0&quot;,&quot;credit&quot;:&quot;joel&quot;,&quot;camera&quot;:&quot;&quot;,&quot;caption&quot;:&quot;&quot;,&quot;created_timestamp&quot;:&quot;1504206443&quot;,&quot;copyright&quot;:&quot;&quot;,&quot;focal_length&quot;:&quot;0&quot;,&quot;iso&quot;:&quot;0&quot;,&quot;shutter_speed&quot;:&quot;0&quot;,&quot;title&quot;:&quot;&quot;,&quot;orientation&quot;:&quot;0&quot;}\" data-image-title=\"modellzusammenfassung\" data-image-description=\"\" data-image-caption=\"\" data-large-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/modellzusammenfassung.jpg\" class=\"alignnone size-medium wp-image-3032\" src=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/modellzusammenfassung-300x237.jpg\" alt=\"\" width=\"300\" height=\"237\" srcset=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/modellzusammenfassung-300x237.jpg 300w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/modellzusammenfassung.jpg 483w\" sizes=\"auto, (max-width: 300px) 100vw, 300px\" \/><\/a><\/strong><\/p>\n<p>The figure below shows the features with their weights which represent the importance for the prediction. You can see that the middle and maximum temperature of the previous day have the biggest impact on the result.<\/p>\n<p><strong><a href=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/pr\u00e4dikatoren.jpg\"><img loading=\"lazy\" decoding=\"async\" data-attachment-id=\"3034\" data-permalink=\"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2017\/08\/31\/wettersave-realizing-weather-forecasts-with-machine-learning\/pradikatoren\/\" data-orig-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/pr\u00e4dikatoren.jpg\" data-orig-size=\"891,422\" data-comments-opened=\"1\" data-image-meta=\"{&quot;aperture&quot;:&quot;0&quot;,&quot;credit&quot;:&quot;joel&quot;,&quot;camera&quot;:&quot;&quot;,&quot;caption&quot;:&quot;&quot;,&quot;created_timestamp&quot;:&quot;1504205594&quot;,&quot;copyright&quot;:&quot;&quot;,&quot;focal_length&quot;:&quot;0&quot;,&quot;iso&quot;:&quot;0&quot;,&quot;shutter_speed&quot;:&quot;0&quot;,&quot;title&quot;:&quot;&quot;,&quot;orientation&quot;:&quot;0&quot;}\" data-image-title=\"pr\u00e4dikatoren\" data-image-description=\"\" data-image-caption=\"\" data-large-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/pr\u00e4dikatoren.jpg\" class=\"alignnone size-medium wp-image-3034\" src=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/pr\u00e4dikatoren-300x142.jpg\" alt=\"\" width=\"300\" height=\"142\" srcset=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/pr\u00e4dikatoren-300x142.jpg 300w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/pr\u00e4dikatoren-768x364.jpg 768w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/pr\u00e4dikatoren.jpg 891w\" sizes=\"auto, (max-width: 300px) 100vw, 300px\" \/><\/a><\/strong><\/p>\n<p>The calculated neural network has one hidden layer with six neurons and is shown below:<\/p>\n<p>&nbsp;<\/p>\n<p><strong><a href=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/Netz.jpg\"><img loading=\"lazy\" decoding=\"async\" data-attachment-id=\"3033\" data-permalink=\"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2017\/08\/31\/wettersave-realizing-weather-forecasts-with-machine-learning\/netz\/\" data-orig-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/Netz.jpg\" data-orig-size=\"794,737\" data-comments-opened=\"1\" data-image-meta=\"{&quot;aperture&quot;:&quot;0&quot;,&quot;credit&quot;:&quot;joel&quot;,&quot;camera&quot;:&quot;&quot;,&quot;caption&quot;:&quot;&quot;,&quot;created_timestamp&quot;:&quot;1504205777&quot;,&quot;copyright&quot;:&quot;&quot;,&quot;focal_length&quot;:&quot;0&quot;,&quot;iso&quot;:&quot;0&quot;,&quot;shutter_speed&quot;:&quot;0&quot;,&quot;title&quot;:&quot;&quot;,&quot;orientation&quot;:&quot;0&quot;}\" data-image-title=\"Netz\" data-image-description=\"\" data-image-caption=\"\" data-large-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/Netz.jpg\" class=\"alignnone size-medium wp-image-3033\" src=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/Netz-300x278.jpg\" alt=\"\" width=\"300\" height=\"278\" srcset=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/Netz-300x278.jpg 300w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/Netz-768x713.jpg 768w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/Netz.jpg 794w\" sizes=\"auto, (max-width: 300px) 100vw, 300px\" \/><\/a><\/strong><\/p>\n<p>To produce the models which use two or four previous days to predict the next day, the datasets had to be adjusted within the SPSS Modeler. Every data set must contain the measurements of the two or four previous days as well as the data of the day for the prediction. For that we had to use operations for choosing, filtering, connecting and ordering data. In the following picture, you can see the stream which uses four previous days for calculating the model.<\/p>\n<p><strong><a href=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day.jpg\"><img loading=\"lazy\" decoding=\"async\" data-attachment-id=\"3030\" data-permalink=\"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2017\/08\/31\/wettersave-realizing-weather-forecasts-with-machine-learning\/4day\/\" data-orig-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day.jpg\" data-orig-size=\"1057,753\" data-comments-opened=\"1\" data-image-meta=\"{&quot;aperture&quot;:&quot;0&quot;,&quot;credit&quot;:&quot;joel&quot;,&quot;camera&quot;:&quot;&quot;,&quot;caption&quot;:&quot;&quot;,&quot;created_timestamp&quot;:&quot;1504206084&quot;,&quot;copyright&quot;:&quot;&quot;,&quot;focal_length&quot;:&quot;0&quot;,&quot;iso&quot;:&quot;0&quot;,&quot;shutter_speed&quot;:&quot;0&quot;,&quot;title&quot;:&quot;&quot;,&quot;orientation&quot;:&quot;0&quot;}\" data-image-title=\"4day\" data-image-description=\"\" data-image-caption=\"\" data-large-file=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day-1024x729.jpg\" class=\"alignnone size-medium wp-image-3030\" src=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day-300x214.jpg\" alt=\"\" width=\"300\" height=\"214\" srcset=\"https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day-300x214.jpg 300w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day-768x547.jpg 768w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day-1024x729.jpg 1024w, https:\/\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/4day.jpg 1057w\" sizes=\"auto, (max-width: 300px) 100vw, 300px\" \/><\/a><\/strong><\/p>\n<p>&nbsp;<\/p>\n<p>&nbsp;<\/p>\n<p>After executing the stream, a model is created (yellow diamond at the right site) that can be used in another stream which is used for calculating the resulting prediction and uploaded at Bluemix.<br \/>\nWe noticed that including the data of two or four previous days as input at the prediction model instead of the data of only one day creates only minimal better results (1 day: 92,4%, 2 days: 92,9%, 4 days: 92,8%).<\/p>\n<p><strong>Server and website<\/strong><\/p>\n<p>After uploading the streams to Watson at Bluemix, we connected all parts with a Node.js server and created the website. As this was only a side aspect of the project, I would not describe this process more detailed.<\/p>\n<p><strong>The Result<\/strong><\/p>\n<p>You can test the web service at wettersave.mybluemix.net. You will see that only dates from 12\/23\/2015 until 6\/20\/2017 are possible to predict as described at the paragraph about the database.<\/p>\n<p>If you are interested in the code, you can find it at the following GitLab repository:<\/p>\n<p>https:\/\/gitlab.mi.hdm-stuttgart.de\/sh245\/cloud<\/p>\n","protected":false},"excerpt":{"rendered":"<p>&nbsp;&nbsp;&nbsp;&nbsp;Introduction Since the internet boom a few years ago companies started to collect and save data in an almost aggressive way. But the huge amounts of data are actually useless if they are not used to gain new information with a higher value. I was always impressed by the way how easy statistical algorithms can [&hellip;]<\/p>\n","protected":false},"author":485,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[1,652],"tags":[],"ppma_author":[735],"class_list":["post-3029","post","type-post","status-publish","format-standard","hentry","category-allgemein","category-artificial-intelligence"],"aioseo_notices":[],"jetpack_featured_media_url":"","jetpack-related-posts":[{"id":1017,"url":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2016\/07\/25\/machine-learning-in-secure-systems\/","url_meta":{"origin":3029,"position":0},"title":"Machine Learning in secure systems","author":"Claudius Messerschmidt","date":"25. July 2016","format":false,"excerpt":"Sadly today's security systems often be hacked and sensitive informations get stolen. To protect a company against cyber-attacks security experts define a \"rule set\" to detect and prevent any attack. This \u201canalyst-driven solutions\u201d are build up from human experts with their domain knowledge. This knowledge is based on experiences and\u2026","rel":"","context":"In &quot;Allgemein&quot;","block_context":{"text":"Allgemein","link":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/category\/allgemein\/"},"img":{"alt_text":"","src":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2016\/07\/Machine_learning_SeSy_robot_landscape.jpg?resize=350%2C200&ssl=1","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2016\/07\/Machine_learning_SeSy_robot_landscape.jpg?resize=350%2C200&ssl=1 1x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2016\/07\/Machine_learning_SeSy_robot_landscape.jpg?resize=525%2C300&ssl=1 1.5x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2016\/07\/Machine_learning_SeSy_robot_landscape.jpg?resize=700%2C400&ssl=1 2x"},"classes":[]},{"id":2615,"url":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2017\/08\/27\/fooling-the-intelligence\/","url_meta":{"origin":3029,"position":1},"title":"FOOLING THE INTELLIGENCE","author":"Jonas Miederer","date":"27. August 2017","format":false,"excerpt":"Adversarial machine learning and its dangers The world is led by machines, humans are subjected to the robot\u2019s rule. Omniscient computer systems hold the control of the world. The newest technology has outpaced human knowledge, while the mankind is powerless in the face of the stronger, faster, better and almighty\u2026","rel":"","context":"In &quot;Allgemein&quot;","block_context":{"text":"Allgemein","link":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/category\/allgemein\/"},"img":{"alt_text":"","src":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/AAEAAQAAAAAAAAxmAAAAJDcyNzkyZjkzLTUzZTEtNGU1ZS04OWYxLWU4NDU5Y2QxOTRjYQ.png?resize=350%2C200&ssl=1","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/AAEAAQAAAAAAAAxmAAAAJDcyNzkyZjkzLTUzZTEtNGU1ZS04OWYxLWU4NDU5Y2QxOTRjYQ.png?resize=350%2C200&ssl=1 1x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/AAEAAQAAAAAAAAxmAAAAJDcyNzkyZjkzLTUzZTEtNGU1ZS04OWYxLWU4NDU5Y2QxOTRjYQ.png?resize=525%2C300&ssl=1 1.5x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/AAEAAQAAAAAAAAxmAAAAJDcyNzkyZjkzLTUzZTEtNGU1ZS04OWYxLWU4NDU5Y2QxOTRjYQ.png?resize=700%2C400&ssl=1 2x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2017\/08\/AAEAAQAAAAAAAAxmAAAAJDcyNzkyZjkzLTUzZTEtNGU1ZS04OWYxLWU4NDU5Y2QxOTRjYQ.png?resize=1050%2C600&ssl=1 3x"},"classes":[]},{"id":10318,"url":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2020\/04\/13\/open-source-batch-and-stream-processing-realtime-analysis-of-big-data\/","url_meta":{"origin":3029,"position":2},"title":"Open Source Batch and Stream Processing: Realtime Analysis of Big Data","author":"Marcel Stolin","date":"13. April 2020","format":false,"excerpt":"Abstract Since the beginning of Big Data, batch processing was the most popular choice for processing large amounts of generated data. These existing processing technologies are not suitable to process the large amount of data we face today. Research works developed a variety of technologies that focus on stream processing.\u2026","rel":"","context":"In &quot;Allgemein&quot;","block_context":{"text":"Allgemein","link":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/category\/allgemein\/"},"img":{"alt_text":"","src":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2023\/08\/mapreduce.jpg?resize=350%2C200&ssl=1","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2023\/08\/mapreduce.jpg?resize=350%2C200&ssl=1 1x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2023\/08\/mapreduce.jpg?resize=525%2C300&ssl=1 1.5x"},"classes":[]},{"id":5262,"url":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2019\/02\/26\/reproducibility-in-ml\/","url_meta":{"origin":3029,"position":3},"title":"Reproducibility in Machine Learning","author":"Pascal Fecht","date":"26. February 2019","format":false,"excerpt":"The rise of Machine Learning has led to changes across all areas of computer science. From a very abstract point of view, heuristics are replaced by black-box machine-learning algorithms providing \"better results\". But how do we actually quantify better results? ML-based solutions tend to focus more on absolute performance improvements\u2026","rel":"","context":"In &quot;Allgemein&quot;","block_context":{"text":"Allgemein","link":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/category\/allgemein\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":23138,"url":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2022\/03\/31\/an-overview-of-large-scale-deep-learning\/","url_meta":{"origin":3029,"position":4},"title":"An overview of Large Scale Deep Learning","author":"mk374","date":"31. March 2022","format":false,"excerpt":"article by Annika Strau\u00df (as426) and Maximilian Kaiser (mk374) Introduction Improving Deep Learning with ULS for superior model training Single Instance Single Device (SISD) Multi Instance Single Device (MISD) Multi Instance Multi Device (MIMD) Single Instance Multi Device (SIMD) Model parallelism Data parallelism Improving ULS and its components with the\u2026","rel":"","context":"In &quot;Allgemein&quot;","block_context":{"text":"Allgemein","link":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/category\/allgemein\/"},"img":{"alt_text":"","src":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2022\/03\/quantum-physics-g1357f44f5_1920-Kopie.jpg?resize=350%2C200&ssl=1","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2022\/03\/quantum-physics-g1357f44f5_1920-Kopie.jpg?resize=350%2C200&ssl=1 1x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2022\/03\/quantum-physics-g1357f44f5_1920-Kopie.jpg?resize=525%2C300&ssl=1 1.5x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2022\/03\/quantum-physics-g1357f44f5_1920-Kopie.jpg?resize=700%2C400&ssl=1 2x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2022\/03\/quantum-physics-g1357f44f5_1920-Kopie.jpg?resize=1050%2C600&ssl=1 3x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2022\/03\/quantum-physics-g1357f44f5_1920-Kopie.jpg?resize=1400%2C800&ssl=1 4x"},"classes":[]},{"id":5523,"url":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/2019\/02\/28\/federated-learning\/","url_meta":{"origin":3029,"position":5},"title":"Federated Learning","author":"Simon Fabian Wolf","date":"28. February 2019","format":false,"excerpt":"The world is enriched daily with the latest and most sophisticated achievements of Artificial Intelligence (AI). But one challenge that all new technologies need to take seriously is training time. With deep neural networks and the computing power available today, it is finally possible to perform the most complex analyses\u2026","rel":"","context":"In &quot;Allgemein&quot;","block_context":{"text":"Allgemein","link":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/category\/allgemein\/"},"img":{"alt_text":"","src":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2023\/08\/Federated_Learning_averaging_sw189.png?resize=350%2C200&ssl=1","width":350,"height":200,"srcset":"https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2023\/08\/Federated_Learning_averaging_sw189.png?resize=350%2C200&ssl=1 1x, https:\/\/i0.wp.com\/blog.mi.hdm-stuttgart.de\/wp-content\/uploads\/2023\/08\/Federated_Learning_averaging_sw189.png?resize=525%2C300&ssl=1 1.5x"},"classes":[]}],"jetpack_sharing_enabled":true,"authors":[{"term_id":735,"user_id":485,"is_guest":0,"slug":"sh245","display_name":"sh245@hdm-stuttgart.de","avatar_url":"https:\/\/secure.gravatar.com\/avatar\/df098cd5d5350dced60ad1ad57fc1550dcdc88d6ee59af99174fbbffe0a43d74?s=96&d=mm&r=g","0":null,"1":"","2":"","3":"","4":"","5":"","6":"","7":"","8":""}],"_links":{"self":[{"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/posts\/3029","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/users\/485"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/comments?post=3029"}],"version-history":[{"count":3,"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/posts\/3029\/revisions"}],"predecessor-version":[{"id":24733,"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/posts\/3029\/revisions\/24733"}],"wp:attachment":[{"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/media?parent=3029"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/categories?post=3029"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/tags?post=3029"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/blog.mi.hdm-stuttgart.de\/index.php\/wp-json\/wp\/v2\/ppma_author?post=3029"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}