The reason they are blackboxes is because they are function approximators with billions of parameters. Theory has not caught up with practical results. This is why you tune hyperparameters (learning rate, number of layers, number of neurons ina layer, etc.) and have multiple iterations of training to get an approximation of the distribution of the inputs. Training is also sensitive to the order of inputs to the network. A network trained on the same training set but in a different order might converge to an entirely different function. This is why you train on the same inputs in random order over multiple episodes to hopefully average out such variations. They are blackboxes simply because you can’t yet prove theoretically the function it has approximated or converged to given the input.
I’ve been using KOReader on my phone now, ever since my Kindle one day decided to be unrecognisable on my computer. Couldn’t find a solution to fix it so it became a glorified paperweight.
The screen real estate is slightly degraded, but fuck if I give Amazon any more of my money. Besides, I get to store epubs as epubs instead of converting to that god awful mobi format.