Robustness of entrusted data-analysis in e-science research collaboration

Abstract

In e-science, a participant may entrust data-analysis to a third party. This analysis may necessitate digital watermarking for security/privacy reasons. When we use watermarking, there is a robustness requirement that the data be kept acceptable. By using a phantom experiment, We demonstrate that the traditional robustness evaluation does not always work in the e-science scenario.