This requires the image processing toolbox (for bwlabel). It also requires that A is sorted (which appears to be the case in your example). Finally, note that the following will also group together [1 2 3 4] since the difference between consecutive numbers is < 2 even the the difference between the extreme is > 2. If you don't want that then you need to explain how the numbers should be grouped in that case.
A=[0.5 1 1.75 2 5 5.25 5.75 6 6.25 9 9.25 9.5 10.25];
assert(issorted(A), 'A must be sorted');
result = diag(bwlabel(abs(A - A.') <= 2, 4)).'
edit: Actually, it can be done easily without the image processing toolbox. The same conditions still apply:
A=[0.5 1 1.75 2 5 5.25 5.75 6 6.25 9 9.25 9.5 10.25];
assert(issorted(A), 'A must be sorted');
result = cumsum([1, diff(A) > 2])
