Full Resolution Image Compression with Recurrent Neural Networks
A neural network is a computer system modeled on the human brain and nervous system. Artificial neural networks are inspired by biological neural networks and are used to estimate and approximate functions that can depend on a large number of inputs that are generally unknown. This article relates to this class because neural networks are like an information network, which we will talk about later in the class. Neural networks help connect the information and help us make a decision at the end or do something else based on the information.
The linked article article talks about how Google uses neural networks and trained computer intelligence to make images smaller than current JPEG compression allows. They did this by using the network to learn compression from a sample of 6 million compressed photos broken into 32×32 pixel pieces. The linked paper shows how the neural network broke down the images into binary code and reconstructed them piece by piece, In the end, this neural network outperformed JPEG compression at most bitrates.