论文部分内容阅读
This paper expounds a data-fitting algorithm for the double-weight neural network,and presents a new algorithm for the system’s power management on the base of that.The double-weight neural network learns knowledge from the past idle periods of the system,and predicts the lengths of the coming idle periods.As a result of that,the system can switch its running states and re- duce the power dissipation according to the predictive values.The results of the experiments prove that this algorithm shows a better performance in increasing the right rate of shutting down and reducing the power consumption than other traditional ones.
This paper expounds a data-fitting algorithm for the double-weight neural network, and presents a new algorithm for the system’s power management on the base of that. The double-weight neural network learns knowledge from the past idle periods of the system, and predicts the lengths of the coming idle periods.As a result of that, the system can switch its running states and re- duce the power dissipation according to the predictive values. The results of the experiments prove that this algorithm shows a better performance in increasing the right rate of shutting down and reducing the power consumption than other traditional ones.