Here I have implemented CAE - Convolution AutoEncoder for the task of feature extraction on the CIFAR-10 Dataset using keras and TenserFlow.
- Contains 60,000 images of 32x32 pixels belonging to 10 mutually exclusive classes: airplane, automobile, bird, cat, deer, dog, frog, horse, ship, and truck.
- Includes 50,000 training images and 10,000 testing images.
- https://paperswithcode.com/dataset/cifar-10
A Convolutional Autoencoder (CAE) is a type of artificial neural network that learns to compress and reconstruct input data (often images) in a compressed, low-dimensional representation.
-
CAEs are effective for feature extraction, as the latent space captures important features of the input data in a low-dimensional form.
-
They can be used for dimensionality reduction, data compression, and anomaly detection.
-
Compared to regular autoencoders, they are specifically designed for images and leverage the spatial structure of the data.
