EdgeServe: A Streaming System for Decentralized Model Serving
This addresses the need for efficient real-time model serving in decentralized environments, though it appears incremental as it builds on existing streaming systems.
The paper tackles the problem of serving machine learning models over continuous data streams by presenting EdgeServe, a distributed streaming system that enables real-time predictions, as demonstrated on tasks like human activity recognition, autonomous driving, and network intrusion detection.
The relevant features for a machine learning task may arrive as one or more continuous streams of data. Serving machine learning models over streams of data creates a number of interesting systems challenges in managing data routing, time-synchronization, and rate control. This paper presents EdgeServe, a distributed streaming system that can serve predictions from machine learning models in real time. We evaluate EdgeServe on three streaming prediction tasks: (1) human activity recognition, (2) autonomous driving, and (3) network intrusion detection.