Melola Discovery

← Melola Discovery

Jurnal SINTA 1

A Low-complexity Complex-valued Activation Function for Fast and Accurate Spectral Domain Convolutional Neural Network

Indonesian Journal of Electical Engineering and Informatics (IJEEI) · Vol. 9 No. 1 (2021) · hlm. 173–184

DOI
10.52549/ijeei.v9i1.2737
ISSN
2089-3272, 2089-3272
Terbit
2021-03-03
Penerbit
IAES Indonesian Section
Bahasa
en

Abstrak

Conventional Convolutional Neural Networks (CNNs), which are realized in spatial domain, exhibit high computational complexity. This results in high resource utilization and memory usage and makes them unsuitable for implementation in resource and energy-constrained embedded systems. A promising approach for low-complexity and high-speed solution is to apply CNN modeled in the spectral domain. One of the main challenges in this approach is the design of activation functions. Some of the proposed solutions perform activation functions in spatial domain, necessitating multiple and computationally expensive spatial-spectral domain switching. On the other hand, recent work on spectral activation functions resulted in very computationally intensive solutions. This paper proposes a complex-valued activation function for spectral domain CNNs that only transmits input values that have positive-valued real or imaginary component. This activation function is computationally inexpensive in both forward and backward propagation and provides sufficient nonlinearity that ensures high classification accuracy. We apply this complex-valued activation function in a LeNet-5 architecture and achieve an accuracy gain of up to 7% for MNIST and 6% for Fashion MNIST dataset, while providing up to 79% and 85% faster inference times, respectively, over state-of-the-art activation functions for spectral domain.

Kata kunci

Jaringan kolaborasi para penulis

5 penulis · 7 tautan kepenulisan bersama

Arahkan kursor atau ketuk sebuah nama untuk menyorot rekan penulisnya. Klik dua kali untuk membuka halaman penulis.

  • University of Technology Malaysia
  • Organisasi berbeda
  • Organisasi sama
  • Afiliasi tidak diketahui
  • Garis lebih tebal = lebih banyak artikel bersama · lingkaran lebih besar = lebih banyak artikel
Tampilkan sebagai tabel
Rekan penulisOrganisasiArtikel bersama
Nurulnajah Mohd ZabidiUniversity of Technology Malaysia1

Afiliasi berasal dari OpenAlex; metadata OAI-PMH jurnal tidak memuatnya. Penulis tanpa ID OpenAlex dicocokkan berdasarkan nama, sehingga dua orang bisa tergabung atau satu orang bisa terpisah.

Sumber metadata: OAI-PMH (metadata jurnal) http://section.iaesonline.com/index.php/IJEEI/oai, pertama di-harvest 8 Nov 2026, terakhir terlihat 8 Nov 2026.
Metadata asli (oai_dc)
<oai_dc:dc xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/  http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
	<dc:title xml:lang="en">A Low-complexity Complex-valued Activation Function for Fast and Accurate Spectral Domain Convolutional Neural Network</dc:title>
	<dc:creator xml:lang="en">Rizvi, Shahriyar Masud</dc:creator>
	<dc:creator xml:lang="en">Ab Rahman, Ab Al-Hadi</dc:creator>
	<dc:creator xml:lang="en">Khalil-Hani, Mohamed</dc:creator>
	<dc:creator xml:lang="en">Ayat, Sayed Omid</dc:creator>
	<dc:subject xml:lang="en">Convolutional Neural Network</dc:subject>
	<dc:subject xml:lang="en">Spectral domain CNN</dc:subject>
	<dc:subject xml:lang="en">Activation functions</dc:subject>
	<dc:subject xml:lang="en">Embedded systems</dc:subject>
	<dc:description xml:lang="en">Conventional Convolutional Neural Networks (CNNs), which are realized in spatial domain, exhibit high computational complexity. This results in high resource utilization and memory usage and makes them unsuitable for implementation in resource and energy-constrained embedded systems. A promising approach for low-complexity and high-speed solution is to apply CNN modeled in the spectral domain. One of the main challenges in this approach is the design of activation functions. Some of the proposed solutions perform activation functions in spatial domain, necessitating multiple and computationally expensive spatial-spectral domain switching. On the other hand, recent work on spectral activation functions resulted in very computationally intensive solutions. This paper proposes a complex-valued activation function for spectral domain CNNs that only transmits input values that have positive-valued real or imaginary component. This activation function is computationally inexpensive in both forward and backward propagation and provides sufficient nonlinearity that ensures high classification accuracy. We apply this complex-valued activation function in a LeNet-5 architecture and achieve an accuracy gain of up to 7% for MNIST and 6% for Fashion MNIST dataset, while providing up to 79% and 85% faster inference times, respectively, over state-of-the-art activation functions for spectral domain.</dc:description>
	<dc:publisher xml:lang="en">IAES Indonesian Section</dc:publisher>
	<dc:date>2021-03-03</dc:date>
	<dc:type>info:eu-repo/semantics/article</dc:type>
	<dc:type>info:eu-repo/semantics/publishedVersion</dc:type>
	<dc:format>application/pdf</dc:format>
	<dc:identifier>https://section.iaesonline.com/index.php/IJEEI/article/view/2737</dc:identifier>
	<dc:identifier>10.52549/ijeei.v9i1.2737</dc:identifier>
	<dc:source xml:lang="en">Indonesian Journal of Electrical Engineering and Informatics (IJEEI); Vol. 9 No. 1: March 2021; 173-184</dc:source>
	<dc:source>2089-3272</dc:source>
	<dc:source>2089-3272</dc:source>
	<dc:language>en</dc:language>
	<dc:relation>https://section.iaesonline.com/index.php/IJEEI/article/view/2737/600</dc:relation>
</oai_dc:dc>