Upload
jeffry-owens
View
213
Download
1
Embed Size (px)
Citation preview
1
Heat Diffusion Classifier on a Graph
Haixuan Yang, Irwin King, Michael R. Lyu The Chinese University of Hong Kong
Group Meeting2006
2
Introduction
Heat Diffusion Model on a Graph
Three Graph Inputs
Connections with Other Models
Experiments
Conclusions and Future Work
Outline
3
IntroductionKondor & Lafferty (NIPS2002)
Construct a diffusion kernel on a graphApply to a large margin classifier
Lafferty & Kondor (JMLR2005)Construct a diffusion kernel on a special manifoldApply to SVM
Belkin & Niyogi (Neural Computation 2003)Reduce dimension by heat kernel and local distance
Tenenbaum et al (Science 2000) Reduce dimension by local distance
4
Introduction
The ideas we inherit Local information
relatively accurate in a nonlinear manifold. Heat diffusion on a manifold
a generalization of the Gaussian density from Euclidean space to manifold.
heat diffuses in the same way as Gaussian density in the ideal case when the manifold is the Euclidean space.
The ideas we think differently Establish the heat diffusion equation directly on a graph
three proposed candidate graphs. Construct a classifier by the solution directly.
5
Heat Diffusion Model on a Graph
Notations
. of period a during neighbor its from receives
, at timeheat that ofamount :)(
. at time nodeat heat the:)(
matrix. weight the:))((
}. to from edgean is thereif :){(
}{
wheregraph, weighteddirectedgiven a :)(
Δtji
ti,j,t,Δ,M
tii,tf
i,jwW=
jii,jE=
.1,2,...,nV
V,E,WG
6
Heat Diffusion Model on a Graph
Assumptions
. differenceheat the toalproportion is ) ( ,t)f(j,t)-f(iΔti,j,t,M
before. described as spaceEuclidean in the
doesit as way same in the to from flowsheat The ij
. period time the toalproportion is ) ( ΔtΔti,j,t,M
Δtt(t)-ff ααΔti,j,t,M ijβ
w-
2ij
))(() (
7
Heat Diffusion Model on a Graph
Solution
Eijij
β
w-
ii
ii
Δtt(t)-ffαetfΔttf
tfΔttf2ij
),(
))(()()(
:as expressed becan )( and )( differenceheat The
)()()(
:formmatrix a as expressed becan It
tαHfΔt
t-fΔttf
)0()0()()()(
:becomesequation above the,0Let
fefetftαHftfdt
d
Δt
HtH
8
Heat Diffusion Model on a Graph
Three candidate graphs KNN Graph
Connect points j and i from j to i if j is one of the K nearest neighbors of i, measured by the Euclidean distance.
SKNN-Graph Choose the smallest K*n/2 undirected edges,
which amounts to K*n directed edges. Minimum Spanning Tree
Choose the subgraph such that It is a tree connecting all vertices; the sum of
weights is minimum among all such trees.
9
Heat Diffusion Model on a Graph Illustration
Manifold KNN Graph SKNN-Graph Minimum Spanning Tree
10
Heat Diffusion Model on a Graph Advantages and disadvantages
KNN Graph Democratic to each node Resulting classifier is a generalization of KNN May not be connected Long edges may exit while short edges are removed
SKNN-Graph Not democratic May not be connected Short edges are more important than long edges
Minimum Spanning Tree Not democratic Long edges may exit while short edges are removed Connection is guaranteed Less parameter Faster in training and testing
11
Heat Diffusion Classifier (HDC)
Choose a graph Compute the heat kernel Compute the heat distribution for
each class according to the initial heat distribution
Classify according to the heat distribution
12
Connections with other models The Parzen window approach (when the window
function takes the normal form) is a special case of the HDC for the KNN and SKNN graphs (whenγis small, K=n-1).
KNN is a special case of the HDC for the KNN graph (whenγis small, 1/β=0).
In Euclidean space, the proposed heat diffusion model for the KNN graph (when K is set to be 2m, 1/β=0) is a generalization of the solution deduced by Finite Difference Method.
Hopefield Model (PNAS, 1982) is the original one which determines class by looking at immediate neighbors. (Thanks to the anonymous reviewer)
13
Experiments Experimental Setup
Experimental Environments Hardware: Nix Dual Intel Xeon
2.2GHz OS: Linux Kernel 2.4.18-27smp
(RedHat 7.3) Developing tool: C
Data Description 3 artificial Data sets and 6
datasets from UCI Comparison
Algorithms: Parzen window
KNNSVM KNN-HSKNN-HMST-H
Results: average of the ten-fold cross validation
DatasetCase
sClasses Variable
Syn-1 100 2 2
Syn-2 100 2 3
Syn-3 200 2 3
Breast-w 683 2 9
Glass 214 6 9
Iono 351 2 34
Iris 150 3 4
Sonar 208 2 60
Vehicle 846 4 18
14
Experiments
Results Dataset SVM KNN PWA KNN-H MST-H SKNN-H
Syn-1 66.0 67.0 80.0 93.0 95.0 95.0
Syn-2 34.0 67.0 83.0 94.0 94.0 89.0
Syn-3 54.0 79.5 92.0 91.0 90.0 92.0
Breast-w 96.8 94.1 96.6 96.9 95.9 99.4
Glass 68.1 61.2 63.5 68.1 68.7 70.5
Iono 93.7 83.2 89.2 96.3 96.3 96.3
Iris 96 97.3 95.3 98.0 92.0 94.7
Sonar 88.5 80.3 53.9 90.9 91.8 94.7
Vehicle 84.8 63.0 66.0 65.5 83.5 66.6
15
Conclusions and Future Work
KNN-H, SKNN-H and MST-H Candidates for the Heat Diffusion Classifier on a
Graph.
Future Work Apply the asymmetric exp{γH} to SVM. Extend the current heat diffusion model further
(from inside). DiffusionRank is a generalization of PageRank