Efficient neural networks for edge devices

Due to limited computation and storage resources of industrial internet things (IoT) edge devices, many emerging intelligent IoT applications based on deep neural networks (DNNs) heavily depend cloud computing for storage. However, faces technical issues in long latency, poor reliability, weak privacy, resulting the need on-device On-device is essential time-critical applications, which require real-time data processing. In this paper, we review three major research areas computation, specifically quantization, pruning, network architecture design. The techniques could enable a DNN model be deployed devices storage, mainly due reduction space complexity. More importantly, these make DNNs applicable devices.