GONE: A generic O(1) NoisE layer for protecting privacy of deep neural networks.
Haibin ZhengJinyin ChenWenchang ShangguanZhaoyan MingXing YangZhijun YangPublished in: Comput. Secur. (2023)
Keyphrases
- neural network
- multi layer
- privacy preservation
- privacy preserving
- pattern recognition
- noise level
- multiple layers
- privacy issues
- protection mechanisms
- single layer
- random noise
- access control
- personal information
- application layer
- self organizing maps
- privacy preserving data mining
- image noise
- data protection
- fuzzy logic
- private information
- noise model
- data security
- sensitive information
- signal to noise ratio
- genetic algorithm
- missing data
- noisy data
- multilayer perceptron
- neural nets
- data privacy
- recurrent neural networks
- differential privacy
- security issues
- neural network model
- gaussian noise
- back propagation
- input data
- domain specific
- inference control
- sensitive data