1: \begin{abstract}

2: In manifold learning, algorithms based on graph Laplacians

3: constructed from data have received considerable attention both in

4: practical applications and theoretical analysis. In particular, the

5: convergence of graph Laplacians obtained from sampled data to

6: certain continuous operators has become an active research topic

7: recently. Most of the existing work has been done under the

8: assumption that the data is sampled from a manifold without boundary

9: or that the functions of interests are evaluated at a point away

10: from the boundary. However, the question of boundary behavior is of

11: considerable practical and theoretical interest. In this paper we

12: provide an analysis of the behavior of graph Laplacians at a point

13: near or on the boundary, discuss their convergence rates and their

14: implications and provide some numerical results. It turns out that

15: while points near the boundary occupy only a small part of the total

16: volume of a manifold, the behavior of  graph Laplacian there has

17: different scaling properties from its behavior elsewhere on the

18: manifold, with global effects on the whole manifold, an observation

19: with potentially important implications for the general problem of

20: learning on manifolds.

21: \end{abstract}