Publication result detail

Multi-objective Evolutionary Neural Architecture Search for Hailo Accelerators

NEPRAŠ, D.; SEKANINA, L.

Original Title

Multi-objective Evolutionary Neural Architecture Search for Hailo Accelerators

English Title

Multi-objective Evolutionary Neural Architecture Search for Hailo Accelerators

Type

Paper in proceedings (conference paper)

Original Abstract

A multi-objective, hardware-aware Neural Architecture Search (NAS) framework is introduced and evaluated for efficient design of convolutional neural networks (CNN) on the Hailo-8L accelerator. The search for CNNs that exhibit good accuracy-latency trade-offs on image classification tasks is conducted using NSGA-II and a single-objective genetic algorithm that imposes a performance constraint. To reduce execution time, we adopt a pretrained supernet, introduce an optimized evolution-training interaction, and develop a latency predictor based on a surrogate model. The framework is primarily intended to construct CNN architectures optimized for Hailo-8L and user-provided applicationspecific datasets. Nevertheless, to facilitate a direct comparison with existing methods, we evaluated it on standard image classification benchmarks, specifically CIFAR-10, CIFAR-100, and ImageNet-100.

English abstract

A multi-objective, hardware-aware Neural Architecture Search (NAS) framework is introduced and evaluated for efficient design of convolutional neural networks (CNN) on the Hailo-8L accelerator. The search for CNNs that exhibit good accuracy-latency trade-offs on image classification tasks is conducted using NSGA-II and a single-objective genetic algorithm that imposes a performance constraint. To reduce execution time, we adopt a pretrained supernet, introduce an optimized evolution-training interaction, and develop a latency predictor based on a surrogate model. The framework is primarily intended to construct CNN architectures optimized for Hailo-8L and user-provided applicationspecific datasets. Nevertheless, to facilitate a direct comparison with existing methods, we evaluated it on standard image classification benchmarks, specifically CIFAR-10, CIFAR-100, and ImageNet-100.

Keywords

Neural architecture search, Hardware accelerator, Genetic algorithm, Inference latency, Image classification

Key words in English

Neural architecture search, Hardware accelerator, Genetic algorithm, Inference latency, Image classification

Authors

NEPRAŠ, D.; SEKANINA, L.

Released

09.05.2026

Publisher

Springer Nature

Location

CHAM

ISBN

978-3-032-23603-6

Book

Applications of Evolutionary Computation, EvoApplications

Periodical

Lecture Notes in Computer Science

Volume

16524

State

Federal Republic of Germany

Pages from

484

Pages to

500

Pages count

17

BibTex

@inproceedings{BUT212657,
  author="David {Nepraš} and Lukáš {Sekanina}",
  title="Multi-objective Evolutionary Neural Architecture Search for Hailo Accelerators",
  booktitle="Applications of Evolutionary Computation, EvoApplications",
  year="2026",
  journal="Lecture Notes in Computer Science",
  volume="16524",
  pages="484--500",
  publisher="Springer Nature",
  address="CHAM",
  doi="10.1007/978-3-032-23604-3",
  isbn="978-3-032-23603-6",
  issn="0302-9743"
}