PROSAC: Provably Safe Certification for Machine Learning Models under Adversarial Attacks.
Ziquan LiuZhuo ZhiIlija BogunovicCarsten Gerner-BeuerleMiguel RodriguesPublished in: CoRR (2024)
Keyphrases
- machine learning models
- spam filtering
- machine learning algorithms
- machine learning approaches
- machine learning
- watermarking scheme
- predictive model
- learning models
- third party
- evolutionary algorithm
- training data
- hidden markov models
- probabilistic model
- machine learning methods
- learning problems
- neural network