Supplement to “ Graph Sparsification Approaches for Laplacian Smoothing ”

نویسندگان

  • Veeranjaneyulu Sadhanala
  • Yu-Xiang Wang
  • Ryan J. Tibshirani
چکیده

This document contains proofs, supplementary details, and supplementary experiments for the paper " Graph Sparsification Approaches for Laplacian Smoothing ". All section numbers, equation numbers, and figure numbers in this supplementary document are preceded by the letter A, to distinguish them from those from the main paper. Part (a). By optimality ofˆθ for problem (5), y − ˆ θ 2 2 + λ ˆ θ T ˜ L ˆ θ ≤ y − ˆ β 2 2 + λ ˆ β T ˜ L ˆ β ≤ y − ˆ β 2 2 + λ(1 +) ˆ β T L ˆ β, where we have used the spectral similarity of L, ˜ L. Rearranging, we see thatˆθ 2 2 − ˆ β 2 2 ≤ 2y T (ˆ θ − ˆ β) + λ(1 +) ˆ β T L ˆ β − λ ˆ θ T ˜ L ˆ θ. Substituting y = ˆ β + y − ˆ β on the right-hand side, and again rearranging, ˆ θ − ˆ β 2 2 ≤ 2(y − ˆ β) T (ˆ θ − ˆ β) + λ(1 +) ˆ β T L ˆ β − λ ˆ θ T ˜ L ˆ θ. Using y − ˆ β = λLˆβ from the stationarity condition for (1), ˆ θ − ˆ β 2 2 ≤ 2λˆθ T L ˆ β − λ(1 −) ˆ β T L ˆ β − λ ˆ θ T ˜ L ˆ θ ≤ 2λL 1/2 ˆ θ 2 L 1/2 ˆ β 2 − λ(1 −) ˆ β T L ˆ β − λ ˆ θ T ˜ L ˆ θ ≤ 2λ √ 1 + ˜ L 1/2 ˆ θ 2 L 1/2 ˆ β 2 − λ(1 −) ˆ β T L ˆ β − λ ˆ θ T ˜ L ˆ θ, where we have again used the spectral similarity of L, ˜ L. Now we examine two cases for last line above. If √ 1 + ˜ L 1/2 ˆ θ 2 ≤ L 1/2 ˆ β 2 , thenˆθ − ˆ β 2 2 ≤ λ(1 +) ˆ β T L ˆ β − λ ˆ θ T ˜ L ˆ θ. If √ 1 + ˜ L 1/2 ˆ θ 2 > L 1/2 ˆ β 2 , thenˆθ − ˆ β 2 2 ≤ λ(1 + 2) ˆ …

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Graph Sparsification Approaches for Laplacian Smoothing

Given a statistical estimation problem where regularization is performed according to the structure of a large, dense graph G, we consider fitting the statistical estimate using a sparsified surrogate graph G̃, which shares the vertices of G but has far fewer edges, and is thus more tractable to work with computationally. We examine three types of sparsification: spectral sparsification, which c...

متن کامل

The Laplacian Polynomial and Kirchhoff Index of the k-th‎ Semi Total Point Graphs

The k-th semi total point graph of a graph G, , ‎is a graph‎ obtained from G by adding k vertices corresponding to each edge and‎ connecting them to the endpoints of edge considered‎. ‎In this paper‎, a formula for Laplacian polynomial of in terms of‎ characteristic and Laplacian polynomials of G is computed‎, ‎where is a connected regular graph‎.The Kirchhoff index of is also computed‎.

متن کامل

Towards Scalable Spectral Clustering via Spectrum-Preserving Sparsification

The eigendeomposition of nearest-neighbor (NN) graph Laplacian matrices is the main computational bottleneck in spectral clustering. In this work, we introduce a highly-scalable, spectrum-preserving graph sparsification algorithm that enables to build ultra-sparse NN (u-NN) graphs with guaranteed preservation of the original graph spectrums, such as the first few eigenvectors of the original gr...

متن کامل

Spectral Sparsification of Graphs

We introduce a new notion of graph sparsification based on spectral similarity of graph Laplacians: spectral sparsification requires that the Laplacian quadratic form of the sparsifier approximate that of the original. This is equivalent to saying that the Laplacian of the sparsifier is a good preconditioner for the Laplacian of the original. We prove that every graph has a spectral sparsifier ...

متن کامل

Incremental Spectral Sparsification for Large-Scale Graph-Based Semi-Supervised Learning

While the harmonic function solution performs well in many semi-supervised learning (SSL) tasks, it is known to scale poorly with the number of samples. Recent successful and scalable methods, such as the eigenfunction method [11] focus on efficiently approximating the whole spectrum of the graph Laplacian constructed from the data. This is in contrast to various subsampling and quantization me...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2016