# Exploring Deep-to-Shallow Transformable Neural Networks for Intelligent Embedded Systems

Source: [Exploring Deep-to-Shallow Transformable Neural Networks for Intelligent Embedded Systems](<https://www.sota2.com/research/paper/694b738e53fdd5c361739a91>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-12-25T05:00:04.588Z

Authors: Xiangzhong Luo, Weichen Liu

Published: 2025-12-17T07:47:10.000Z

## Abstract

Thanks to the evolving network depth, convolutional neural networks (CNNs) have achieved remarkable success across various embedded scenarios, paving the way for ubiquitous embedded intelligence. Despite its promise, the evolving network depth comes at the cost of degraded hardware efficiency. In contrast to deep networks, shallow networks can deliver superior hardware efficiency but often suffer from inferior accuracy. To address this dilemma, we propose Double-Win NAS, a novel deep-to-shallow transformable neural architecture search (NAS) paradigm tailored for resource-constrained intelligent embedded systems. Specifically, Double-Win NAS strives to automatically explore deep networks to first win strong accuracy, which are then equivalently transformed into their shallow counterparts to further win strong hardware efficiency. In addition to search, we also propose two enhanced training techniques, including hybrid transformable training towards better training accuracy and arbitrary-resolution elastic training towards enabling natural network elasticity across arbitrary input resolutions. Extensive experimental results on two popular intelligent embedded systems (i.e., NVIDIA Jetson AGX Xavier and NVIDIA Jetson Nano) and two representative large-scale datasets (i.e., ImageNet and ImageNet-100) clearly demonstrate the superiority of Double-Win NAS over previous state-of-the-art NAS approaches.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Neural Architecture Search on ImageNet (train val)](<https://www.sota2.com/research/sota/neural-architecture-search-on-imagenet-train-val.md>): Neural Architecture Search; ImageNet (train val); Search Cost (GPU-days): 0.6.

- [Image Classification on CIFAR-10 (Standard Performance)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-standard-performance.md>): Image Classification; CIFAR-10; Latency (ms): 1.04.

- [Object Detection on COCO 2017](<https://www.sota2.com/research/sota/object-detection-on-coco-2017.md>): Object Detection; COCO 2017; AP (Box): 21.6.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2512.19731>)

[PDF](<https://arxiv.org/pdf/2512.19731.pdf>)