NPAS: A Compiler-aware Framework of Unified Network Pruning and Architecture Search for Beyond Real-Time Mobile Acceleration

TL;DR

This paper introduces NPAS, a compiler-aware framework that unifies network pruning and architecture search, optimized for mobile acceleration, achieving significant speedups and high accuracy on ImageNet inference.

Contribution

It presents a novel unified framework combining model pruning, NAS, and compiler optimization, with a meta-modeling approach for efficient search in mobile deployment.

Findings

Achieves inference times of 3.9-6.7ms on ImageNet with high accuracy.

Outperforms prior methods in mobile DNN acceleration.

Supports various DNN architectures and pruning schemes.

Abstract

With the increasing demand to efficiently deploy DNNs on mobile edge devices, it becomes much more important to reduce unnecessary computation and increase the execution speed. Prior methods towards this goal, including model compression and network architecture search (NAS), are largely performed independently and do not fully consider compiler-level optimizations which is a must-do for mobile acceleration. In this work, we first propose (i) a general category of fine-grained structured pruning applicable to various DNN layers, and (ii) a comprehensive, compiler automatic code generation framework supporting different DNNs and different pruning schemes, which bridge the gap of model compression and NAS. We further propose NPAS, a compiler-aware unified network pruning, and architecture search. To deal with large search space, we propose a meta-modeling procedure based on reinforcement learning with fast evaluation and Bayesian optimization, ensuring the total number of training epochs comparable with representative NAS frameworks. Our framework achieves 6.7ms, 5.9ms, 3.9ms ImageNet inference times with 78.2%, 75% (MobileNet-V3 level), and 71% (MobileNet-V2 level) Top-1 accuracy respectively on an off-the-shelf mobile phone, consistently outperforming prior work.