Tiered Pruning for Efficient Differentialble Inference-Aware Neural Architecture Search

TL;DR

This paper introduces three innovative pruning methods for differentiable neural architecture search, significantly enhancing inference efficiency and achieving state-of-the-art results on ImageNet and COCO benchmarks.

Contribution

It presents Prunode, a stochastic bi-path building block, and novel pruning algorithms for blocks and layers within the SuperNet, advancing inference-aware neural architecture search.

Findings

Achieves new Pareto optimality for inference latency and accuracy on ImageNet.

Outperforms GPUNet and EfficientNet in object detection latency and mAP.

Reduces computational complexity with O(1) memory for inner hidden dimension search.

Abstract

We propose three novel pruning techniques to improve the cost and results of inference-aware Differentiable Neural Architecture Search (DNAS). First, we introduce Prunode, a stochastic bi-path building block for DNAS, which can search over inner hidden dimensions with O(1) memory and compute complexity. Second, we present an algorithm for pruning blocks within a stochastic layer of the SuperNet during the search. Third, we describe a novel technique for pruning unnecessary stochastic layers during the search. The optimized models resulting from the search are called PruNet and establishes a new state-of-the-art Pareto frontier for NVIDIA V100 in terms of inference latency for ImageNet Top-1 image classification accuracy. PruNet as a backbone also outperforms GPUNet and EfficientNet on the COCO object detection task on inference latency relative to mean Average Precision (mAP).