| Abstract (EN) |
This research aimed to (1) investigate the performance of a Convolutional Neural
Network (CNN) for image classification using the CIFAR-10 dataset, and (2) compare the
performance of the CNN model with baseline techniques for image classification on the CIFAR10 dataset using deep learning approaches. The methodology involved designing a CNN
architecture consisting of convolutional layers, pooling layers, and fully connected layers.
Model optimization techniques were applied, including the use of activation functions,
overfitting reduction methods, and hyperparameter tuning. The performance of the models
was evaluated using Accuracy and Loss metrics. The results indicated that (1) the
Convolutional Neural Network effectively learned image features from the CIFAR-10 dataset,
and (2) in the comparative performance analysis, the CNN achieved the highest performance
with an accuracy of 0.7015, followed by HOG + SVM (Accuracy = 0.6208) and LBP + k-NN
(Accuracy = 0.1012), respectively. The findings demonstrate that the Convolutional Neural
Network has strong potential for practical applications in image processing tasks. |
Creative Commons License