Let b1 and b2 be basis vectors for the plane and let x0 be a point in the plane. Then, for example,
x0(:) + b1(:)*randn(1,N) + b2(:)*abs(randn(1,N));
will result in N random points on one side of the dividing line x0+t*b1 (the one passing through x0 and parallel to b1), while
x0(:) + b1(:)*randn(1,M) - b2(:)*abs(randn(1,M));
will result in M points on the opposite side.