Particle swarm optimization for convolutional autoencoders.

Photo by Alina Grubnyak on Unsplash

Convolutional neural networks offer a specialized architecture to multi-dimensional data like images, video, or multi-dimensional time series, just to mention a few. In a convolutional neural network, a convolutional kernel is applied to the input which generates a feature score on which we apply the activation function. Repeating this process to successive layers results in a deep…