Convolutional Neural Networks are widely used in various real-world applications due to their exceptional performance. To further enhance this effectiveness, many approaches utilize spatial attentions ...

Abstract: We propose a convolutional autoencoder neural network for image classification in YCbCr color space to reduce computational complexity. We first learned local image features from image ...

SAE + DBSCAN produces high-quality clusters but requires significantly more training time. Visual inspection confirmed cluster cohesion and successful feature learning.