|
Autoencoder Applications for Enhancing Spice Classification Performance under Limited Dataset Conditions |
|---|---|
| รหัสดีโอไอ | |
| Creator | Evan Tanuwijaya |
| Title | Autoencoder Applications for Enhancing Spice Classification Performance under Limited Dataset Conditions |
| Contributor | Andreas Lim, Nathanael Suryanto, Joey Wiryawan |
| Publisher | Faculty of Informatics, Mahasarakham University |
| Publication Year | 2569 |
| Journal Title | Journal of Applied Informatics and Technology |
| Journal Vol. | 8 |
| Journal No. | 1 |
| Page no. | 256450 |
| Keyword | Autoencoder, Convolution Neural Network, Limited Dataset, Spices |
| URL Website | https://ph01.tci-thaijo.org/index.php/jait |
| Website title | Journal of Applied Informatics and Technology |
| ISSN | 3088-1803 |
| Abstract | Indonesia is home to a vast array of spices, known for their health benefitsand essential roles in culinary traditions. However, widespread knowledgeof these spices remains limited. This study explores the use of autoencodertechnology to address the challenges of classifying spices based on theirintricate shapes and colors, especially when faced with the limitations ofa small dataset. By carefully adjusting hyperparameters such as learningrate, optimization function, input size, and epochs, the autoencoder is op-timized to extract distinctive features necessary for accurate classification.A comparison with convolutional neural networks (CNNs) using transferlearning shows that the autoencoder achieves similar accuracy levels, whileconsistently avoiding issues of overfitting or underfitting. With precisionand recall rates nearing 0.97, the autoencoder demonstrates its abilityto compress image data and identify key features, making it an effectivesolution for spice classification in constrained data environments. |
