Critical Overview of Privacy-Preserving Learning in Vector Autoregressive Models for Energy Forecasting
Cooperation between different data owners may lead to an improvement of forecast skill by, for example, taking advantage of spatio-temporal dependencies in geographically distributed renewable energy time series. Due to business competitive factors and personal data protection, these data owners might be unwilling to share their information, which increases the interest in collaborative privacy-preserving forecasting. This paper analyses the state-of-the-art and unveils several shortcomings of existing methods in guaranteeing data privacy when employing Vector Autoregressive (VAR) models. Mathematical proofs and numerical analysis are conducted to evaluate existing privacy-preserving methods divided into three categories: data transformation, secure multi-party computations, and decomposition methods. The analysis shows that state-of-the-art techniques have limitations in preserving data privacy, such as a trade-off between privacy and the correct estimation of model coefficients, while iterative processes in which intermediate results are shared can be exploited so that the original data can be inferred after some iterations.
READ FULL TEXT