Design of Embedded Software Based on Rough Set and Neural Network

At the time of designing embedded software, too many variables in the original requirement specifications may cause the design work very difficult. This paper tries to use the knowledge reduction in the rough set to simplify the design of embedded software. We first get raw data from requirement specifications or use cases and then turn raw data into rough set by the process of the discretization. Next, extract knowledge (or decision rules) from the rough set by the knowledge reduction. Finally, employing the neural network theory learned these decision rules from the embedded software. In brief, rough set theory is able to extract the decision rules from the raw data, and then the neural network is capable of learning these decision rules to help development team works out.