Resource-Aware Edge-Based Stream Analytics

Understanding how machine learning (ML) algorithms can be used for stream processing on edge devices remains an important challenge. Such ML represented as operators and dynamically adapted based the resources which they are hosted. Deploying often focuses carrying out inference edge, while model development takes place a cloud data center. In this article, we describe TinyMOA, modified version of open-source massive online analytics library processing, that deployed across both local remote using Parsl Kafka systems. Using experimental testbed, demonstrate stream-processing configured resource hosted, discuss subsequent implications edge-based