BottleNet: A Deep Learning Architecture for Intelligent Mobile Cloud Computing Services
This addresses the challenge of efficient mobile cloud computing for users needing real-time AI services, representing a strong specific gain rather than a foundational breakthrough.
The paper tackles the problem of high latency and energy consumption in mobile deep learning by introducing BottleNet, a deep learning architecture that splits networks between mobile devices and the cloud, achieving on average 30x improvement in latency and 40x improvement in energy consumption with negligible accuracy loss.
Recent studies have shown the latency and energy consumption of deep neural networks can be significantly improved by splitting the network between the mobile device and cloud. This paper introduces a new deep learning architecture, called BottleNet, for reducing the feature size needed to be sent to the cloud. Furthermore, we propose a training method for compensating for the potential accuracy loss due to the lossy compression of features before transmitting them to the cloud. BottleNet achieves on average 30x improvement in end-to-end latency and 40x improvement in mobile energy consumption compared to the cloud-only approach with negligible accuracy loss.