Abstract:
Machine learning as a service (MLaaS) has brought much convenience to our daily lives recently. However, the fact that the service is provided through cloud raises privac...Show MoreMetadata
Abstract:
Machine learning as a service (MLaaS) has brought much convenience to our daily lives recently. However, the fact that the service is provided through cloud raises privacy leakage issues. In this work we propose the compressive privacy generative adversarial network (CPGAN), a data-driven adversarial learning framework for generating compressing representations that retain utility comparable to state-of-the-art, with the additional feature of defending against reconstruction attack. This is achieved by applying adversarial learning scheme to the design of compression network (privatizer), whose utility/privacy performances are evaluated by the utility classifier and the adversary reconstructor, respectively. Experimental results demonstrate that CPGAN achieves better utility/privacy trade-off in comparison with the previous work, and is applicable to real-world large datasets.
Published in: IEEE Transactions on Information Forensics and Security ( Volume: 15)