Quantifying identifiability to choose and audit epsilon in differentially private deep learning.
Daniel BernauGünther EiblPhilip-William GrassalHannah KellerFlorian KerschbaumPublished in: Proc. VLDB Endow. (2021)
Keyphrases
- deep learning
- differentially private
- differential privacy
- unsupervised learning
- machine learning
- unsupervised feature learning
- intrusion detection
- mental models
- data mining techniques
- active learning
- contingency tables
- weakly supervised
- expectation maximization
- data model
- privacy preserving
- learning strategies
- object recognition
- image segmentation
- search engine
- information retrieval