Secure decision tree inference using Bloom filters
University of New Brunswick
Cloud computing allows model providers to distribute machine learning models at scale without purchasing dedicated hardware for model hosting. However, when hosting their models in the cloud, model providers may be forced to disclose private model details. Due to the time and monetary investments associated with model training, model providers may be reluctant to host their models due to these privacy concerns. To combat these issues, several privacy preserving decision tree schemes have been proposed which ensure the privacy of the decision tree models, the client query, and the final classification of the model. However, most existing schemes require significant communication or computational overhead. In this work, we propose a privacy preserving scheme for decision tree inference, which uses Bloom filters to hide the original decision tree structure while maintaining reliable classification results. Our scheme’s security and performance are verified through rigorous testing and analysis.