Read on blog in more details.
An autoencoder is the combination of an encoder function that converts the input data into a different representation and a decoder function that converts the new representation back into the original format. It's a data compression algorithm where the compression and decompression functions are
- Data-specific,
- Lossy, and
- Learned automatically from examples rather than engineered by a human.
As it's data specific and lossy, it's not good for image compression in general. The fact that autoencoders are data-specific which makes them generally impractical for real-world data compression problems. But there's a hope, future advances might change this. I find it interesting, though it's not good enoguh and also very poor performance compared to othr compression algorithm like JPEG, MPEG etc. Check out this keras blog post regarding on this issue.
And also fllowing, in case of you're interested too.
output:
