Variance Reduction for Inverse Trace Estimation via Random Spanning Forests
read the original abstract
The trace $\tr(q(\ma{L} + q\ma{I})^{-1})$, where $\ma{L}$ is a symmetric diagonally dominant matrix, is the quantity of interest in some machine learning problems. However, its direct computation is impractical if the matrix size is large. State-of-the-art methods include Hutchinson's estimator combined with iterative solvers, as well as the estimator based on random spanning forests (a random process on graphs). In this work, we show two ways of improving the forest-based estimator via well-known variance reduction techniques, namely control variates and stratified sampling. Implementing these techniques is easy, and provides substantial variance reduction, yielding comparable or better performance relative to state-of-the-art algorithms.
This paper has not been read by Pith yet.
Forward citations
Cited by 2 Pith papers
-
Fast Estimation for Forest Matrix of Signed Graphs
Presents signed forest matrix theorem and GSCF/FMDE algorithms achieving O(n) forest generation and O(ln) diagonal estimation for signed graphs up to 20M nodes.
-
Efficient Computation for Diagonal of Forest Matrix via Variance-Reduced Forest Sampling
New variance-reduced sampling algorithms SCF, SCFV, and SCFV+ compute forest matrix diagonals more efficiently than Laplacian solvers with error guarantees and linear time in nodes for both undirected and directed graphs.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.