| 1. | Sankarshan Damle; Aleksei Triastcyn; Boi Faltings; Sujit Gujar: Differentially Private Multi-Agent Constraint Optimization. The Second AAAI Workshop on Privacy-Preserving Artificial Intelligence (PPAI-21), 2021. (Type: Workshop | Abstract | Links | BibTeX | Tags: differential privacy) @workshop{Damle2021b, title = {Differentially Private Multi-Agent Constraint Optimization}, author = {Sankarshan Damle and Aleksei Triastcyn and Boi Faltings and Sujit Gujar}, editor = {Ferdinando Fioretto}, url = {https://ppai21.github.io/files/38-paper.pdf}, year = {2021}, date = {2021-02-08}, booktitle = {The Second AAAI Workshop on Privacy-Preserving Artificial Intelligence (PPAI-21)}, abstract = {Several optimization scenarios involve multiple agents that desire to protect the privacy of their preferences. There ex- ist distributed algorithms for constraint optimization that pro- vide improved privacy protection through secure multiparty computation. However, it comes at the expense of high com- putational complexity and does not constitute a rigorous pri- vacy guarantee, as the result of the computation itself may compromise agents’ preferences. In this work, we show how to achieve differential privacy through randomization of the solving process. In particular, we present P-Gibbs, which adapts the SD-Gibbs algorithm to obtain differential privacy guarantees with much higher computational efficiency. Ex- periments on graph coloring and meeting scheduling show the algorithm’s privacy-performance trade-off compared to variants with uniform sampling and the SD-Gibbs algorithm.}, keywords = {differential privacy}, pubstate = {published}, tppubtype = {workshop} } Several optimization scenarios involve multiple agents that desire to protect the privacy of their preferences. There ex- ist distributed algorithms for constraint optimization that pro- vide improved privacy protection through secure multiparty computation. However, it comes at the expense of high com- putational complexity and does not constitute a rigorous pri- vacy guarantee, as the result of the computation itself may compromise agents’ preferences. In this work, we show how to achieve differential privacy through randomization of the solving process. In particular, we present P-Gibbs, which adapts the SD-Gibbs algorithm to obtain differential privacy guarantees with much higher computational efficiency. Ex- periments on graph coloring and meeting scheduling show the algorithm’s privacy-performance trade-off compared to variants with uniform sampling and the SD-Gibbs algorithm. |
| 1. | : Differentially Private Multi-Agent Constraint Optimization. The Second AAAI Workshop on Privacy-Preserving Artificial Intelligence (PPAI-21), 2021. |