Ekya: Continuous Learning of Video Analytics Models on Edge Compute Servers

TL;DR

Ekya is a system that enables continuous learning for video analytics models on edge servers, balancing inference and retraining to adapt to data drift efficiently while conserving resources.

Contribution

Ekya introduces a novel approach to support inference and retraining simultaneously on edge servers, optimizing accuracy and resource usage through a micro-profiler.

Findings

Ekya achieves 29% higher accuracy than baseline.

Ekya uses 4x less GPU resources to match baseline accuracy.

The system effectively manages the tradeoff between inference and retraining.

Abstract

Video analytics applications use edge compute servers for the analytics of the videos (for bandwidth and privacy). Compressed models that are deployed on the edge servers for inference suffer from data drift, where the live video data diverges from the training data. Continuous learning handles data drift by periodically retraining the models on new data. Our work addresses the challenge of jointly supporting inference and retraining tasks on edge servers, which requires navigating the fundamental tradeoff between the retrained model's accuracy and the inference accuracy. Our solution Ekya balances this tradeoff across multiple models and uses a micro-profiler to identify the models that will benefit the most by retraining. Ekya's accuracy gain compared to a baseline scheduler is 29% higher, and the baseline requires 4x more GPU resources to achieve the same accuracy as Ekya.