Alibaba Cloud Developer

Jun 20, 2018 · Mobile Development

How to Supercharge Mobile Deep Learning: Model Compression & Engine Optimizations

This article explains how to overcome the performance, size, memory, and compatibility challenges of deploying deep‑learning inference engines on mobile devices by jointly optimizing model compression and engine implementation, covering speed tricks, cache‑friendly coding, multithreading, sparsity, quantization, NEON intrinsics, package size reduction, memory pooling, and reliability techniques.