I have a input stream 3D data that comes every time frame. Each point is defined by 3D vector of x,y,z. There is a evaluation function [say f(x)] that computes if the point at time t is valid or invalid. Any point that is marked valid is assign a weight using normal distribution.
I am bit confused how to exactly weight this points according to the history window. That is if points remains in the history window for 5 time frame (5 is max) than it must have very high weight while new added points must have a moderate weight value.
My question is how can I define a weighting function to do this weighting for each point that is valid.
At the moment I defined the weighting function as follows which I am not sure if it is correct.
exp(-1 + (1/5-t)
where 0 <= t <5. But the weights seems not so efficient.