1: \begin{abstract}

2: We study how the supporting hyperplanes produced by the projection

3: process can complement the method of alternating projections and its

4: variants for the convex set intersection problem. For the problem

5: of finding the closest point in the intersection of closed convex

6: sets, we propose an algorithm that, like Dykstra's algorithm, converges

7: strongly in a Hilbert space. Moreover, this algorithm converges in

8: finitely many iterations when the closed convex sets are cones in

9: $\mathbb{R}^{n}$ satisfying an alignment condition. Next, we propose

10: modifications of the alternating projection algorithm, and prove its

11: convergence. The algorithm converges superlinearly in $\mathbb{R}^{n}$

12: under some nice conditions. Under a conical condition, the convergence

13: can be finite. Lastly, we discuss the case where the intersection

14: of the sets is empty.

15: \end{abstract}

16: