Activation functions play a vital role in the training dynamics and generalization performance of deep learning models. This study presents a comparative analysis of ten widely used activation functions—ReLU, Sigmoid, Tanh, ELU, SELU, Softplus, Softsign, Swish, GELU, and a custom spline-based function—within a unified convolutional neural network (CNN) architecture. All models were trained and evaluated on the CIFAR-10 dataset under identical experimental settings, including fixed learning rate, batch size, number of epochs, and architecture configuration.
| Mualliflar | Primbetov, Abbaz, Akbarov, Navruz |
|---|---|
| Jurnal | Al-Farg'oniy avlodlari |
| Nashriyot | ТАТУ Фарғона филиали |
| ISSN | 2181-4252 |
| Nashr sanasi | 2025-10-03 |
| Son | 3 |
| Betlar | 80-84 |
| Til | Ingliz |
| Ilmiy soha | Texnika |
| Indeksga qo‘shilgan | 2026-08-26 |
Activation Functions, Spline Activation, Convolutional Neural Networks, CIFAR-10, Image Classification, ReLU, Swish, Deep Learning
| APA | Primbetov, Abbaz, & Akbarov, Navruz (2025). A COMPARATIVE STUDY OF ACTIVATION FUNCTIONS IN DEEP LEARNING MODELS. Al-Farg'oniy avlodlari, (3), 80-84. https://al-fargoniy.uz/index.php/journal/article/view/893 |
|---|---|
| GOST (OAK) | Primbetov, Abbaz, Akbarov, Navruz A COMPARATIVE STUDY OF ACTIVATION FUNCTIONS IN DEEP LEARNING MODELS // Al-Farg'oniy avlodlari. – 2025. – № 3. – P. 80-84. |
| MLA | Primbetov, Abbaz, and Akbarov, Navruz. “A COMPARATIVE STUDY OF ACTIVATION FUNCTIONS IN DEEP LEARNING MODELS.” Al-Farg'oniy avlodlari, (3), 80-84, 2025. https://al-fargoniy.uz/index.php/journal/article/view/893 |
| Chicago | Primbetov, Abbaz, and Akbarov, Navruz. “A COMPARATIVE STUDY OF ACTIVATION FUNCTIONS IN DEEP LEARNING MODELS.” Al-Farg'oniy avlodlari, (3), 80-84 (2025). https://al-fargoniy.uz/index.php/journal/article/view/893 |
| Harvard | Primbetov, Abbaz, and Akbarov, Navruz (2025) ‘A COMPARATIVE STUDY OF ACTIVATION FUNCTIONS IN DEEP LEARNING MODELS’, Al-Farg'oniy avlodlari, (3), 80-84. https://al-fargoniy.uz/index.php/journal/article/view/893 |
| BibTeX | @article{primbetovabbaz2025,
title = {A COMPARATIVE STUDY OF ACTIVATION FUNCTIONS IN DEEP LEARNING MODELS},
author = {Primbetov, Abbaz and Akbarov, Navruz},
journal = {Al-Farg'oniy avlodlari},
year = {2025},
number = {3},
pages = {80-84},
url = {https://al-fargoniy.uz/index.php/journal/article/view/893}
} |
Boshqa maqola uchun iqtibos tayyorlash
The solutions of two-parameter singularly perturbed boundary value problems usually include two boundary layers. Because of the presence of these layers, conventional numerical methods are considered inefficient for…
Abstract: Heating, Ventilation, and Air Conditioning (HVAC) systems account for about 40% of building energy use. Their performance is strongly affected by uncertain external factors such as weather and stochastic…
The increase in the creation of fake web pages by attackers is leading to a sharp increase in cyberattacks. Attackers use these fake web sites to advertise products to Internet users, distribute malicious programs, or…
This paper addresses the problem of managing multi-parameter data flows in monitoring and treatment of livestock wastewater based on sensor measurements. A sensor network was deployed to collect key water quality…
This article discusses methods for modeling and analyzing satellite communication channel characteristics. Simulations were conducted using MATLAB/Simulink, NS-3, and STK tools based on AWGN, Rayleigh, Rician, and…
This study investigates the causes of atmospheric pollution from motor vehicles and explores methods for reducing the concentration of harmful substances in the air. Based on the analysis of data from various sources…
This article addresses the problem of constructing numerical solutions of hydrodynamic equations in the form of nonlinear partial differential equations describing the three-dimensional velocity of atmospheric air…
In this study, digital image processing and a convolutional neural network (CNN) were integrated to automatically detect mitral stenosis based on 32 echocardiogram videos. By applying data augmentation, the training…
This study analyzes geometry and short-time hydration of two wheat varieties from Uzbekistan, “Asr” and “Alekseevich.” Grain dimensions were measured by digital caliper, microstructure examined microscopically, and…
In this article, the bending process of a rectangular plate under external load is studied both theoretically and practically. The deformation process is based on the theory of elasticity and involves a structural…
Al-Farg'oniy avlodlari — barcha maqolalar · 2025-yil arxivi
Deepfake technology has become a growing threat to information integrity and digital trust. This study examines how accurately humans can identify deepfake videos through an empirical survey conducted among 492 students…
Deepfake technology, powered by artificial intelligence, enables the manipulation of media content in multiple ways, including identity swapping, facial expression alteration, attribute modification, background…
Deepfakes pose a serious threat to media authenticity and public trust. This paper proposes a hybrid deep learning model combining ResNeXt and LSTM to detect deepfakes by capturing both spatial and temporal…
Deepfake technology, driven by advancements in artificial intelligence (AI) and deep learning, facilitates the creation of hyper-realistic synthetic media—encompassing images, videos, and audio. While this innovation…
This article examines compression as a meaning-oriented strategy in sign language translation and interpreting. Compression is understood not as arbitrary shortening, but as the selective reformulation of a source…
This study investigates the impact of Edge Computing on bandwidth utilization in Internet of Things (IoT)–driven Nigerian telecom networks. The rapid growth of IoT devices has significantly increased data traffic…
This article analyzes the scientific, pedagogical, and methodological foundations for the effective integration of national upbringing factors with modern global competencies in the context of continuous education…
This article addresses the problem of improving an adaptive traffic signal control algorithm at urban intersections by taking into account the dynamic characteristics of traffic flow. Fixed-time control under conditions…
Maqolada paxta xomashyosini yirik iflosliklardan tozalash jarayonida arrali baraban va kolosnikli panjara o‘rtasidagi o‘zaro ta’sirni takomillashtirish masalasi ko‘rib chiqilgan. Mavjud tozalagichlarda qo‘llaniladigan…
Maqolada paxta xomashyosini yirik va mayda iflosliklardan tozalash samaradorligini oshirishga qaratilgan ilmiy tadqiqotlar tizimlashtirilib, konstruktiv va texnologik omillarning o‘zaro ta’siri qiyosiy tahlil qilindi…