1: \begin{abstract}

2: 

3: GROUSE (Grassmannian Rank-One Update Subspace

4: Estimation)~\cite{Balzano10b} is an incremental algorithm for

5: identifying a subspace of $\R^n$ from a sequence of vectors in this

6: subspace, where only a subset of components of each vector is revealed

7: at each iteration. Recent analysis~\cite{grouseconverge} has shown

8: that GROUSE converges locally at an expected linear rate, under

9: certain assumptions. GROUSE has a similar flavor to the incremental

10: singular value decomposition algorithm~\cite{Bunch78}, which updates

11: the SVD of a matrix following addition of a single column. In this

12: paper, we modify the incremental SVD approach to handle missing data,

13: and demonstrate that this modified approach is equivalent to GROUSE,

14: for a certain choice of an algorithmic parameter.

15: 

16: %% The GROUSE (Grassmannian Rank-One Update Subspace Estimation)

17: %% algorithm for low-rank matrix completion~\cite{Balzano10b}

18: %% implements stochastic gradient in order to minimize a data fit cost

19: %% function over all subspaces of a given dimension. Because the

20: %% minimization is over all subspaces (i.e., constrained to the

21: %% Grassmannian), it is non-convex. However, in experiments GROUSE

22: %% enjoys repeated success when the true data are actually low-rank or

23: %% when there is a small amount of noise. It has recently been shown

24: %% that GROUSE has a linear convergence rate in a local region of the

25: %% global solution~\cite{grouseconverge}. In our pursuit of a deeper

26: %% understanding, we wish to relate the GROUSE algorithm with the

27: %% Singular Value Decomposition, another algorithm which minimizes a

28: %% non-convex cost (again, a data-fit cost function over all possible

29: %% subspaces) but has a provable polynomial time guarantees in seeking

30: %% the global minimum. In this paper, we show that GROUSE is in fact

31: %% very closely related to the incremental Singular Value

32: %% Decomposition~\cite{Bunch78}.

33: 

34: \end{abstract}

35: