Model Zehirlenmesi
Attacking the behaviour of the model by adding deliberately malicious examples to education data.
Model poisoning can allow attackers to place hidden backdoors or bias in models trained especially with large datasets collected from the internet.