Neural networks are generally assumed to be non-invertible functions. That said, something like an auto-encoder can learn to decode the encoded data back to its

Neural networks are generally assumed to be non-invertible functions. That said, something like an auto-encoder can learn to decode the encoded data back to its original form, with loss.