It's home-made and hurts bad AI...
Failed attempt at building home-made AI poison pill. But AI became a paintbrush for art. You can find some works here - https://gallery.so/neco.
In a way I'm treating the process of modeling AI itself as an art.
- Various color filters made by controling number of training epochs.
- Color contrast effects created by swapping out different activation functions.
- Pixelation control done by placing 2D convolutional layers at the right places.
Realistically, Data poisoning is a nuanced technique that involves -
- Duplicating data to make a model think data distribution is different that what it really is
- Perturbing data with random noise to effect what a model learns per data point
- Injecting false labels
- Swapping out real data with fake data during transmission
- And more.
To make a model learn a distribution over a random noise would only be effective if one were to attempt to learn about the pseudo-randomness that a particular machine that generated poisoned data might be using.