what is the 'NSMethod','exhaustive' k-nearest neighbor classifier

Hello.

I am using the k-nearest neighbor classifier as the below code.

Could you explain what 'NSMethod','exhaustive' are?

Is it related to distance metric learning? (It learns a metric that pulls the neighbor candidates near, while pushes near data from different classes out of the target neighbors margin.) I got a high accuracy than I expected.

clear all
close all


for i = 2:30:750
    X = csvread('kth_optical_only.csv');
    Y = csvread('kth_optical_only_class1.csv');
    X = X(:,1:i);
    Mdl = fitcknn(X,Y,'NumNeighbors',3,...
    'NSMethod','exhaustive','Distance','cosine',...
    'Standardize',1);
    rng(1); % For reproducibility
    CVKNNMdl = crossval(Mdl, 'KFold', 5);
    classAccuracy(i) = 100 - kfoldLoss(CVKNNMdl, 'LossFun', 'ClassifError')*100;
end

Answer 1

[Divya Gaddipati]

Hi,

NSMethod is the Nearest neighbor search method, specified as either 'kdtree' or 'exhaustive'.

• 'kdtree' — Creates and uses a Kd-tree to find nearest neighbors.
• 'exhaustive' — Uses the exhaustive search algorithm. When predicting the class of a new point xnew, the algorithm computes the distance values from all points in X to xnew to find nearest neighbors.

For information on exhaustive and kdtree based searching, you can refer to the following links:

https://www.mathworks.com/help/stats/exhaustivesearcher.html

https://www.mathworks.com/help/stats/kdtreesearcher.html

Hope this helps!

[Kong]

Thank so you much!