Performance Investigation on Image Classification Methods for CIFAR-10

This study investigates the performance of different methods for image classification tasks using the CIFAR-10 dataset. Four neural networks, namely Convolutional Neural Networks (CNNs), Artificial Neural Networks (ANNs), Deep Neural Networks (DNNs), and Long Short-Term Memory (LSTM) networks, were compared based on their precision, recall, training epochs, and memory utilization. Results were analyzed and presented comprehensively. The findings reveal that CNNs outperform the other networks, demonstrating superior memory efficiency, requiring fewer training epochs, and achieving a balanced precision-recall trade-off. While DNNs exhibit high precision, they demand more memory and display lower recall rates. Conversely, LSTM networks show comparable precision and recall scores to CNNs but require more training epochs and memory. The study concludes that CNNs offer the most efficient approach for image classification tasks on the CIFAR-10 dataset, providing valuable insight into selecting methods for similar applications.