What is soft vote in indigenous peoples?

Question

I understand how k-nearest neighbors (KNN) work, but I am not familiar with the term soft voting. What is soft vote on KNN and how does it work compared to standard KNN voting?

A simple example of comparing two voting schemes would be useful, and a reference to the Matlab implementation would be a nice bonus.

thanks

Josh

Answer 1

After some reading, I found that soft voting simply puts gausses at each of the points (training examples) that were voted on.

Usually we just vote for training examples that are closest in the function space, usually adding them to the voices of the closest neighbor (s). Instead, soft voting simply uses the Gaussian probability of all case studies as a voting score and accumulates corresponding votes based on each score. This simply provides a more reliable voting scheme, as it is more aware of relative distances, especially in higher dimensional spaces.

See Mitchell et al. "Soft" voting scheme for the closest neighbors, 2001.

For an example of where it was used, see Agarwal et al. Restoration of a three-dimensional human pose from monocular images, 2005