English

MOGNET: A Mux-residual quantized Network leveraging Online-Generated weights

Machine Learning 2025-01-17 v1 Hardware Architecture

Abstract

This paper presents a compact model architecture called MOGNET, compatible with a resource-limited hardware. MOGNET uses a streamlined Convolutional factorization block based on a combination of 2 point-wise (1x1) convolutions with a group-wise convolution in-between. To further limit the overall model size and reduce the on-chip required memory, the second point-wise convolution's parameters are on-line generated by a Cellular Automaton structure. In addition, MOGNET enables the use of low-precision weights and activations, by taking advantage of a Multiplexer mechanism with a proper Bitshift rescaling for integrating residual paths without increasing the hardware-related complexity. To efficiently train this model we also introduce a novel weight ternarization method favoring the balance between quantized levels. Experimental results show that given tiny memory budget (sub-2Mb), MOGNET can achieve higher accuracy with a clear gap up to 1% at a similar or even lower model size compared to recent state-of-the-art methods.

Keywords

Cite

@article{arxiv.2501.09531,
  title  = {MOGNET: A Mux-residual quantized Network leveraging Online-Generated weights},
  author = {Van Thien Nguyen and William Guicquero and Gilles Sicard},
  journal= {arXiv preprint arXiv:2501.09531},
  year   = {2025}
}

Comments

Published at IEEE AICAS 2022

R2 v1 2026-06-28T21:08:19.185Z