The Influence of Parameter Optimization of VGGNet on Model Performance in Terms of Classification Layers

DOI

10.2991/978-94-6463-540-9_89How to use a DOI?

Keywords

Machine learning; VGGNet; parameter adjustment

Abstract

This paper aims to explore the effect of parameter adjustment in classification layers of VGGNet. It provides suitable amounts of parameters for VGGNet with FC and FCN layers, which are available for reference. In the research, FER13 dataset, which contains gray-scaled images with shape of 48 x 48 with one channel is divided into training, validation and testing dataset. During the procedure of data processing, resizing, color jitter and flipping is adopted to the images to enhance training, and then the images are put into data loaders. As for the architecture of the model, for the reason of the limitation of time and computing capacity, the VGGNet is simplified. In detail, its width and height of convolutional layers are reduced. Also, for the classification layers, FC layers as well as FCN layers with different kernels are adopted. With Adam as optimizer and cross entropy as loss function, the accuracy of each model is tested and compared after training of 20 epochs. Experimental results show the suitable amounts of parameters with which the model has best performance. Also, the results indicate that FCN layers with smaller kernels have better performance than those with larger kernels.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Yizhen He
PY  - 2024
DA  - 2024/10/16
TI  - The Influence of Parameter Optimization of VGGNet on Model Performance in Terms of Classification Layers
BT  - Proceedings of the 2024 2nd International Conference on Image, Algorithms and Artificial Intelligence (ICIAAI 2024)
PB  - Atlantis Press
SP  - 892
EP  - 900
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-540-9_89
DO  - 10.2991/978-94-6463-540-9_89
ID  - He2024
ER  -