Many AI algorithms require huge amounts of data to train, and progress in deep learning and neural networks also adds considerable compute strain to current systems. With the profusion of smaller, lower-powered devices on the edge we also see a huge management and data transfer challenge and the perfect venue for collecting unprecedented learning sets. However, our infrastructure is rarely ready for this.
In this talk I will discuss how a number of open source technologies like NiFi, Spark and TensorFlow can be used together and orchestrated in a massively distributed and cloud centric architecture to do machine learning on edge devices, as well as deep, and informed learning at the core.
Required audience experience: Your background should be technical.
Objective of the talk: The talk will aim at providing practical examples of such applications, including automatic processing of video data across industries like transport, retail and law enforcement as well as local predictive maintenance for the industrial internet.
Keywords: AI, Hadoop, NiFi, Spark, TensorFlow, retail, law enforcement, transport, opensource