The orchestration of <scp>Machine Learning</scp> frameworks with data streams and <scp>GPU</scp> acceleration in <scp>Kafka‐ML</scp> : A deep‐learning performance comparative

Machine Learning (ML) applications need large volumes of data to train their models so that they can make high-quality predictions. Given digital revolution enablers such as the Internet Things (IoT) and Industry 4.0, this information is generated in quantities terms continuous streams not static datasets it case with most AI (Artificial Intelligence) frameworks. Kafka-ML a novel open-source framework allows complete management ML/AI pipelines through streams. In article, we present new features for framework, support well-known PyTorch, well GPU acceleration at different points along pipeline. This pipeline will be described by taking real 4.0 use Petrochemical Industry. Finally, comprehensive evaluation state-of-the-art deep learning carried out demonstrate feasibility platform.