You have to define "similar" more strictly! Here I have take the norm as the similarity measure, but there are other alternatives
% some data
q = randi(5,10,4)
h = randi(5,1,4)
% q(2,:) = h % uncomment to check that minRow will be this row of q
% engine
simFun = @(x) norm(x-h,2) % similarity measure between x and h
simM = arrayfun(@(k) simFun(q(k,:)),1:size(q,1)) % loop over each row of q
[minVal, minRow] = min(simM)
You could put all this in a single line (which is harder to understand and debug).
