Secure Data Publishing of Private Trajectory in Edge Computing of IoT

Algorithm 1

Gradient descent with differential privacy for one epoch.

	Input: input data , output data , the loss function , the data size , the learning rate , the privacy parameters , the gradient norm bound , total epochs , and the current epoch .
(1)	, ,
(2)	for in do
(3)	for do
(4)	compute gradient
(5)	clip gradient
(6)	add noise
(7)
(8)	end for
(9)	end for