Search Results
Kafka-ML: Connecting the data stream with ML/AI frameworks
2022, Martín, Cristian, Langendoerfer, Peter, Zarrin, Pouya Soltani, Díaz, Manuel, Rubio, Bartolomé
Machine Learning (ML) and Artificial Intelligence (AI) depend on data sources to train, improve, and make predictions through their algorithms. With the digital revolution and current paradigms like the Internet of Things, this information is turning from static data to continuous data streams. However, most of the ML/AI frameworks used nowadays are not fully prepared for this revolution. In this paper, we propose Kafka-ML, a novel and open-source framework that enables the management of ML/AI pipelines through data streams. Kafka-ML provides an accessible and user-friendly Web user interface where users can easily define ML models, to then train, evaluate, and deploy them for inferences. Kafka-ML itself and the components it deploys are fully managed through containerization technologies, which ensure their portability, easy distribution, and other features such as fault-tolerance and high availability. Finally, a novel approach has been introduced to manage and reuse data streams, which may eliminate the need for data storage or file systems.
Bridging the Gap Between (AI-) Services and Their Application in Research and Clinical Settings Through Interoperability: the OMI-Protocol
2024-02, Sigle, Stefan, Werner, Patrick, Schweizer, Simon, Caldeira, Liliana, Hosch, René, Dyrba, Martin, Fegeler, Christian, Sigle, Stefan, Werner, Patrick, Schweizer, Simon, Caldeira, Liliana, Hosch, René, Dyrba, Martin, Fegeler, Christian, Grönke, Ana, Seletkov, Dmitrii, Kotter, Elmar, Nensa, Felix, Wehrle, Julius, Kaufmes, Kevin, Scherer, Lucas, Nolden, Marco, Boeker, Martin, Schmidt, Marvin, Pelka, Obioma, Braren, Rickmer, Stump, Shura-Roman, Graetz, Teresa, Pogarell, Tobias, Susetzky, Tobias, Wieland, Tobias, Parmar, Vicky, Wang, Yuanbin
Artificial Intelligence (AI) in research and clinical contexts is transforming the areas of medical and life sciences permanently. Aspects like findability, accessibility, interoperability, and reusability are often neglected for AI-based inference services. The Open Medical Inference (OMI) protocol aims to support remote inference by addressing the aforementioned aspects. Key component of the proposed protocol is an interoperable registry for remote inference services, which addresses the issue of findability for algorithms. It is complemented by information on how to invoke services remotely. Together, these components lay the basis for the implementation of distributed inference services beyond organizational borders. The OMI protocol considers prior work for aspects like data representation and transmission standards wherever possible. Based on Business Process Modeling of prototypical use cases for the service registry and common inference processes, a generic information model for remote services was inferred. Based on this model, FHIR resources were identified to represent AI-based services. The OMI protocol is first introduced using AI-services in radiology but is designed to be generalizable to other application domains as well. It provides an accessible, open specification as blueprint for the introduction and implementation of remote inference services.