{"id":27890,"date":"2025-10-02T10:31:50","date_gmt":"2025-10-02T08:31:50","guid":{"rendered":"https:\/\/wp.eurestools.eu\/key_achievement\/the-cognitive-framework-a-unified-platform-for-scalable-and-trustworthy-ai-ml-lifecycle-management\/"},"modified":"2025-10-02T10:31:50","modified_gmt":"2025-10-02T08:31:50","slug":"the-cognitive-framework-a-unified-platform-for-scalable-and-trustworthy-ai-ml-lifecycle-management","status":"publish","type":"key_achievement","link":"https:\/\/wp.eurestools.eu\/de\/key_achievement\/the-cognitive-framework-a-unified-platform-for-scalable-and-trustworthy-ai-ml-lifecycle-management\/","title":{"rendered":"Der kognitive Rahmen: Eine einheitliche Plattform f\u00fcr skalierbares und vertrauensw\u00fcrdiges KI\/ML-Lebenszyklusmanagement"},"content":{"rendered":"","protected":false},"featured_media":0,"template":"","class_list":["post-27890","key_achievement","type-key_achievement","status-publish","hentry"],"acf":{"logo":"https:\/\/smart-networks.europa.eu\/wp-content\/uploads\/2025\/08\/verge.png","diagram":"https:\/\/smart-networks.europa.eu\/wp-content\/uploads\/advanced-cf7-upload\/VERGE-cognitive-framework-integrating-AI4Edge-and-SPT4AI-pipelines2025053001.png","top10":"","project_name":"VERGE","ka_number":"2","description":"The Cognitive Framework (CF) is a software asset developed in VERGE to support AI\/ML model developers throughout all stages of the model lifecycle, providing a unified, scalable, and compliance-ready software platform. Current ML Operations (MLOps) tools often operate in silos, focusing on specific tasks such as model training, deployment, or monitoring, and therefore lack the cohesion and automation required to manage ML systems at scale. The CF addresses this gap by integrating key features of existing tools into a unified library and incorporating additional functionalities designed to support the intelligent, trustworthy AI solutions developed in VERGE.\nA key design component is the Distributed Knowledge Base (DKB), which implements a unified ontological knowledge base to store and share data across the distributed edge continuum. The DKB serves as a central repository, storing registered AI\/ML models, training datasets, metadata, and infrastructure metrics collected by the observability layer for online training or inference. A novel feature is the recommender system, enabling developers to discover existing models and datasets in the CF that meet specific performance needs.\nWithin the VERGE lifecycle, a first prototype of the DKB and recommender has been developed. Furthermore, the capabilities of the CF have also been exploited to build and manage diverse AI\/ML pipelines in VERGE, including: 1) a split computing solution using deep reinforcement learning to dynamically partition DNN computation between user devices and the edge; 2) a forecasting model combining attention-based time-series methods with temporal clustering to enhance generalization and workload prediction; and 3) the integration of causal-based algorithms and formal verification tools into CF pipelines.\nOverall, the CF holds significant exploitation potential beyond VERGE, with defined blueprints for future enhancements toward higher levels of cognition in upcoming developments","references":"[1] E. Kartsakli (editor), \"Final report on VERGE Edge4AI design,\"\u009d Deliverable D2.2 of VERGE project, April, 2025","category":"CAT-1: Significant Technology Development","sub_categories":"AI\/ML;Cloud Native;Trustworthiness;","call":"Call 1"},"_links":{"self":[{"href":"https:\/\/wp.eurestools.eu\/de\/wp-json\/wp\/v2\/key_achievement\/27890","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/wp.eurestools.eu\/de\/wp-json\/wp\/v2\/key_achievement"}],"about":[{"href":"https:\/\/wp.eurestools.eu\/de\/wp-json\/wp\/v2\/types\/key_achievement"}],"wp:attachment":[{"href":"https:\/\/wp.eurestools.eu\/de\/wp-json\/wp\/v2\/media?parent=27890"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}