PrototypeNAS: Rapid Design of Deep Neural Networks for Microcontroller Units

Mark Deutel; Simon Geis; Axel Plinge

cs.AIMar 16, 2026

PrototypeNAS: Rapid Design of Deep Neural Networks for Microcontroller Units

Authors: Mark Deutel, Simon Geis, Axel Plinge

Organizations: Fraunhofer Institute for Integrated Circuits, Fraunhofer IIS, Germany

Abstract

Enabling efficient deep neural network (DNN) inference on edge devices with different hardware constraints is a challenging task that typically requires DNN architectures to be specialized for each device separately. To avoid the huge manual effort, one can use neural architecture search (NAS). However, many existing NAS methods are resource-intensive and time-consuming because they require the training of many different DNNs from scratch. Furthermore, they do not take the resource constraints of the target system into account. To address these shortcomings, we propose PrototypeNAS, a zero-shot NAS method to accelerate and automate the selection, compression, and specialization of DNNs to different target microcontroller units (MCUs). We propose a novel three-step search method that decouples DNN design and specialization from DNN training for a given target platform. First, we present a novel search space that not only cuts out smaller DNNs from a single large architecture, but instead combines the structural optimization of multiple architecture types, as well as optimization of their pruning and quantization configurations. Second, we explore the use of an ensemble of zero-shot proxies during optimization instead of a single one. Third, we propose the use of Hypervolume subset selection to distill DNN architectures from the Pareto front of the multi-objective optimization that represent the most meaningful tradeoffs between accuracy and FLOPs. We evaluate the effectiveness of PrototypeNAS on 12 different datasets in three different tasks: image classification, time series classification, and object detection. Our results demonstrate that PrototypeNAS is able to identify DNN models within minutes that are small enough to be deployed on off-the-shelf MCUs and still achieve accuracies comparable to the performance of large DNN models.

Figures & tables

Figure 1: Schematic of the three step pipeline of PrototypeNAS.

Optimization	Hyperparameter	Type	Range/Values
architecture	baseline architecture	categorical	task dependent
structural	group depth $i$	categorical	$\left[0,1,2,3\right]$
structural	kernel & stride $i$	categorical	$\left[\left[3,2\right],\left[3,1\right],\left[5,2\right],\left[5,1\right],\left[7,2\right],\left[7,1\right]\right]$
size	width multiplier	continuous	$\left[0.1,1.0\right]$
size	pruning sparsity $i$	continuous	$\left[0.1,0.9\right]$

Table 1: Search space

X

of the MOO ,

\forall i\in\{0,1,2,3\}

resulting in 14 tunable hyperparameters in total.

Figure 2: The proposed search space and how a DNN prototype can be created from it. Each baseline architecture consists of repeatable superblocks, i.e., a predefined pattern of layers, organized in the search space into four groups to be optimized separately.

Panel of Figure 3 — Figure 3: Optimization results. The top-5 DNNs selected from the Pareto front by HSS are marked with red circles.

Optim. Index	210	283	190	311	237
Architecture	MBedNet	MBedNet	SqueezeNet	SqueezeNet	MBedNet
Quantized Test Accuracy [%] ( PTQ )
CIFAR10	$91.8$	$92.5$	$90.3$	$90.0$	93.7
CIFAR100	$67.4$	$70.5$	$66.5$	$63.7$	72.3
ArxPhotos314	$90.8$	$95.4$	97.4	$79.5$	$93.6$
Flowers	$88.9$	$91.6$	$82.9$	$86.3$	93.3

Table 2: Test accuracy of the trained, pruned, and quantized DNNs found by PrototypeNAS for the image datasets. Latency and energy was measured on an iMXRT1062 Cortex-M7 MCU using an external Joulescope energy analyzer.

Optim. Index	224	226	73	273	347
Architecture	MBedNet	MBedNet	Inception	MBedNet	Inception
Quantized Test Accuracy [%] ( QAT )
BitBrain	$83.7$	$87.2$	$81.7$	88.0	$78.9$
Mafaulda	$96.8$	$98.3$	$97.4$	$97.8$	98.5
Daliac	$95.9$	$97.1$	97.2	$97.0$	$96.1$
Compute [MFLOPs]	$36.3$	$43.2$	101.5	123.5	215.3

Table 3: Test accuracy of the trained, pruned, and quantized DNNs found by PrototypeNAS for the time series datasets. Latency and energy was measured on an iMXRT1062 Cortex-M7 MCU using an external Joulescope energy analyzer.

Dataset	MeCo	NASWOT	SNIP	ZiCo	FLOPs
CIFAR10	$0.0$	$0.0$	$-0.2$	0.6	$0.0$
CIFAR100	$0.0$	$0.0$	$-0.2$	0.6	$0.0$
ArxPhotos314	$0.0$	$0.0$	0.2	$-0.2$	$0.0$
Flowers	$0.2$	$0.2$	$-0.4$	0.4	$0.2$
Birds	$0.0$	$0.0$	$-0.2$	0.6	$0.0$
Pets	$0.1$	$-0.4$	$0.2$	0.8	$-0.3$

Table 4: Kendall’s

\tau

scores for the datasets and results shown in Tables 2 and 3 .

Panel of Figure 4 — Figure 4: Kendall’s $\tau$ scores between the four zero-shot proxies for the image and time series classification tasks.

Figure 5: Tradeoff between

mAP^{50}

and FLOPs of the MBedNet YOLO architectures designed by PrototypeNAS.

Panel of Figure 6 — Figure 6: Comparison of PrototypeNAS with TinyNAS and NATS-Bench for CIFAR10.

Panel of Figure 7 — Figure 7: Performance and emission comparison of PrototypeNAS and randomly selected models on CIFAR10.

Explore similar work

Aug 10, 2026

ENAS: An Efficient Hardware-Aware Neural Architecture Search Framework for TinyML on Resource-Constrained Microcontrollers

We present ENAS, a hardware-aware Neural Architecture Search (NAS) framework that combines a static feasibility check, a cell-based search space supporting standard, depthwise-separable, and bottleneck blocks with optional skip connections, and a three-stage hybrid search strategy (random

\rightarrow

top-

K

\rightarrow

mutation) with persistent cross-run caching. Unlike many existing NAS frameworks that rely on GPU acceleration, ENAS is designed to operate efficiently without requiring GPUs, making it suitable for resource-constrained development environments. We evaluate ENAS on two TinyML benchmarks, Visual Wake Words and Melanoma Cancer, across eight microcontrollers with memory footprints ranging from 20,KB to 1,MB SRAM and nine input image resolutions. Our experimental results show that ENAS achieves mean search-time speedups of

2.41{\times}

and

1.70{\times}

on the Visual Wake Words and Melanoma Cancer datasets, respectively, while maintaining competitive test accuracy compared with the recent NanoNAS framework. A measured resource analysis further shows that ENAS-selected models use substantially lower peak activation RAM, the binding constraint for microcontroller deployment at matched accuracy. Additionally, ENAS achieves

79.4\%

test accuracy on an STM32H743-based microcontroller, outperforming the greedy CPU-only baseline by

2.6

percentage points. We release the ENAS framework as open-source at: https://github.com/EdgeIntelligenceLab/ENAS

Mohd Moin Khan, Naman Srivastava, Pandarasamy Arjunan

Indian Institute of Science (IISc), Bengaluru 560012, India

Hardware-Aware NAS Neural Architecture Search Tiny ML

Ask PDF

Jun 15, 2026

An affordable hardware-aware neural architecture search for deploying convolutional neural networks on ultra-low-power computing platforms

Hardware-aware neural architecture search (HW-NAS) allows the integration of Convolutional Neural Networks (CNNs) in microcontrollers devices by automatically designing neural architectures that can fit prearranged hardware constraints. However, state-of-the-art HW-NAS target high-performance microcontrollers, whose power consumption does not meet sensing nodes requirements. This work presents a HW-NAS generating tiny CNNs that can run on ultra-low-power microcontrollers, featuring a lightweight search procedure enabling its execution even on embedded devices. Empirical results on three well-known benchmarks for tiny computer vision proved that the proposed HW-NAS was able to generate tiny CNNs while preserving state-of-the-art classification accuracy.

Andrea Mattia Garavagno, Edoardo Ragusa, Antonio Frisoli +1

Dept. of Electrical, Electronic, Telecommunications Engineering and Naval Architecture (DITEN), University of Genoa, Genoa, Italy · Department of Excellence in Robotics & AI and Institute of Mechanical Intelligence, Scuola Superiore Sant’Anna, 56127 Pisa, Italy

Efficient Neural Network Inference Tiny ML Hardware-Aware NAS Convolutional Neural Networks Edge ML Lightweight CNNs+1

Ask PDF

May 20, 2026

AutoMCU: Feasibility-First MCU Neural Network Customization via LLM-based Multi-Agent Systems

Deploying neural networks on microcontroller units (MCUs) is critical for edge intelligence but remains challenging due to tight memory, storage, and computation constraints. Existing approaches, such as model compression and hardware-aware neural architecture search (HW-NAS), often depend on proxy metrics, incur high search cost, and do not fully bridge the gap between architecture design and verified deployment. This paper presents AutoMCU, a feasibility-first large language model (LLM)-based multi-agent system for automated neural network customization under MCU constraints. Given natural-language task requirements and hardware specifications, AutoMCU iteratively generates structured architecture candidates, filters infeasible designs through vendor toolchain feedback before training, evaluates feasible models under a controlled protocol, and verifies deployability through backend-grounded deployment analysis. AutoMCU includes two key mechanisms: 1) hardware-in-the-loop architecture generation for early elimination of undeployable candidates under RAM and Flash constraints, and 2) state-isolated multi-agent scheduling for stable coordination of proposal, training, evaluation, and deployment stages. Experiments on CIFAR-10 and CIFAR-100 under strict MCU constraints show that AutoMCU achieves competitive accuracy while reducing customization time to about 1--2 hours, compared with hundreds of GPU hours for representative MCU-oriented HW-NAS baselines. Comparisons with ColabNAS and the LLM-based NAS method GENIUS on NAS-Bench-201 further demonstrate the effectiveness and stability of AutoMCU. Real-device deployments on multiple STM32 microcontrollers validate its practical applicability to MCU-scale edge intelligence.

Penglin Dai, Zijie Zhou, Xincao Xu +3

School of Computing and Artificial Intelligence, Southwest Jiaotong University, Chengdu 611756, China · Shenzhen Institute for Advanced Study, University of Electronic Science and Technology of China, Shenzhen 518110, China · School of Computer Science and Engineering, Northeastern University, Shenyang 110819, China

Efficient Neural Network Inference Hardware-Aware NAS Edge ML Automated Machine Learning

Ask PDF