Research on quadratic programming homomorphic encrypted gradient descent algorithm for an information system
The homomorphic encryption method can realize the function of analyzing or operating encrypted data without leaking the content.However,this method has a heavy computational overhead and has been very hard to achieve in practice within the control domain..In order to solve this problem,we first calculated and analyzed the applicability of the gradient descent algorithm in solving quadratic programming in the homomorphic encryption method,reduced the limitation of the multiplication depth of the homomorphic encryption circuit on the iteration of the gradient descent algorithm,and quantified the prototype example.Lay the foundation for further research.Secondly,the choice of gradient descent and accel-erated gradient descent methods was weighed and evaluated,which opened up a path for the engineering application of ho-momorphic encryption technology.The adopted CKKS scheme allows the program to achieve convergence by selecting an appropriate step size,directly demonstrating the feasibility of the homomorphic encryption gradient descent algorithm.
NETL
NSTL
万方数据
