Ekya: Continuous Learning of Video Analytics Models on Edge Compute Servers
This work is significant for organizations deploying video analytics on edge servers, as it provides a method to maintain model accuracy in the face of data drift with efficient resource utilization.
This paper addresses the challenge of data drift in video analytics models deployed on edge servers by continuously retraining them. Their solution, Ekya, balances the tradeoff between retrained model accuracy and inference accuracy, achieving a 29% higher accuracy gain compared to a baseline scheduler while requiring 4x fewer GPU resources for the same accuracy.
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.