The increasing concerns about data privacy and security drives the emergence of a new field of studying privacy-preserving machine learning from isolated data sources . We propose to protect the decision path by the efficient additivelyhomomorphic encryption method, which allows the disclosure of feature names and thus makes the federated decision trees interpretable . The advantages of Fed-EINI will be demonstrated through theoretical analysis and extensivenumerical results . Experiments show that the inference efficiency is improved by over $50\%$ in average in average. Experiments showed that the infraction efficiency ofFed-EinI is improvedby over . over . $50/€ in average . Inventor: Fed-INI is an efficient and interpretable framework for federated . decision tree models with only one round of multi-party communication. We shall compute the candidate sets of leaf nodes in parallel in parallel, followed by securelycomputing the only leaf node in the . data at each party in parallel to the . only leaf nodes

Author(s) : Xiaolin Chen, Shuai Zhou, Kai Yang, Hao Fan, Zejin Feng, Zhong Chen, Hu Wang, Yongji Wang

Links : PDF - Abstract

Code :
Coursera

Keywords : decision - fed - average - federated - data -

Leave a Reply

Your email address will not be published. Required fields are marked *