{"id":25466,"date":"2021-09-14T00:00:00","date_gmt":"2024-09-13T08:48:48","guid":{"rendered":"https:\/\/wordpress-583806-4798031.cloudwaysapps.com\/three-big-ai-news-for-it-architects-and-uml-or-sysml-users\/"},"modified":"2024-11-15T16:21:30","modified_gmt":"2024-11-15T15:21:30","slug":"three-big-ai-news-for-it-architects-and-uml-or-sysml-users","status":"publish","type":"post","link":"https:\/\/informator.se\/en\/three-big-ai-news-for-it-architects-and-uml-or-sysml-users\/","title":{"rendered":"Three Big AI News for IT-Architects, and UML or SysML Users"},"content":{"rendered":"\n

What I wrote about the ways AI affects architects<\/a>, four years ago, fits into the architectural landscape of 2021 as well. That said, this summer I\u2019m adding one of my \u201cand vice versa-s\u201d, to my quote from AI2 (The Allen Institute of AI, founded by Paul Allen of Microsoft) \u201cNever<\/strong> mistake a clear view for a short distance\u201d… Nor <\/strong>mistake a haze for a long distance, especially not in architecture. The current growth rate of AI in health & life sciences<\/a> (while 4,000 new publications are added daily to the 30 million on PubMed.gov) and in several other domains is breathtaking.  <\/em><\/p>\n\n\n\n

1.<\/strong>  December: prize-winning DeepMind AlphaFold<\/a> was hailed after solving the protein-folding problem. Only 7 months later, in July 2021: DeepMind already used it to predict the shapes of nearly every protein in the human body, as well as the shapes of hundreds of thousands of other proteins. DeepMind will predict and release the structures for another 100 million<\/em><\/a> in the next few months<\/em>. Which means? Well, practically all proteins known to science.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

2.  <\/strong>Sahlgrenska<\/a> Science Park in Gotheburg, county councils, and local AI spinoffs collaborate<\/a> on medical-Swedish<\/a> NLP models; yet another health care<\/a>\/life sciences<\/a> initiative in current applied AI in the Nordics<\/a>, including XAI<\/a>, GAN<\/a>s, Federated ML<\/a>, NLP libraries, and more.<\/p>\n\n\n\n

3.<\/strong>  September 2021: Object Management Group<\/a>, the open consortium behind IT-model standards (e.g. the UML,<\/a> SysML<\/a>, MDA, CORBA), launches its Artificial Intelligence Platform Task Force AIPTF<\/a>. They will standardize foundational capabilities of AI, freeing AI suppliers as well as users from interoperability and interchange limitations with respect to models, languages and data. They\u2019ll focus on modeling of AI artifacts, neural networks, ML & DL, knowledge representations, NLP, security, and they will welcome new members.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

UML<\/em><\/strong> Multiple Generalization (akin  but not the same as: ) AI<\/strong> random Decision Forest, principle <\/em>Diagrams from Informator courses (1) Avancerad Modellering m UML (2) AI Architecture and ML<\/p>\n\n\n\n

Most likely, the AIPTF strengthens the global XAI efforts to close the gap between accuracy<\/em> and explainability<\/em>\/transparency <\/em>of recent AI tech such as deep neural networks (in extreme cases, you might say the Epic Split<\/a> not far from the <\/em>viral Volvo video\u2026)<\/p>\n\n\n\n

Fingers crossed (or legs gradually less apart ?)<\/p>\n\n\n\n

Milan’s courses at Informator:<\/h4>\n\n\n\n