Random forest algorithm under differential privacy
| Title | Random forest algorithm under differential privacy |
| Publication Type | Conference Paper |
| Year of Publication | 2017 |
| Authors | Li, Z., Li, S. |
| Conference Name | 2017 IEEE 17th International Conference on Communication Technology (ICCT) |
| Date Published | oct |
| ISBN Number | 978-1-5090-3944-9 |
| Keywords | Classification algorithms, classification process, composability, data privacy, data privacy disclosure, Decision Tree, Decision trees, Differential privacy, DPRF-gini, Forestry, gini index, Human Behavior, Indexes, original algorithm, privacy, pubcrawl, Random Forest, random forest algorithm, Resiliency, Scalability |
| Abstract | Trying to solve the risk of data privacy disclosure in classification process, a Random Forest algorithm under differential privacy named DPRF-gini is proposed in the paper. In the process of building decision tree, the algorithm first disturbed the process of feature selection and attribute partition by using exponential mechanism, and then meet the requirement of differential privacy by adding Laplace noise to the leaf node. Compared with the original algorithm, Empirical results show that protection of data privacy is further enhanced while the accuracy of the algorithm is slightly reduced. |
| URL | https://ieeexplore.ieee.org/document/8359960 |
| DOI | 10.1109/ICCT.2017.8359960 |
| Citation Key | li_random_2017 |