SlideShare ist ein Scribd-Unternehmen logo
1 von 20
Downloaden Sie, um offline zu lesen
ABRA: APPROXIMATING BETWEENNESS CENTRALITY
IN STATIC AND DYNAMIC GRAPHS WITH
RADEMACHER AVERAGES
Matteo Riondata and Eli Upfal
22nd ACM SIGKDD Conference, August 2016
1
Murata Lab - Paper reading seminar
Presented by: Kaushalya Madhawa
(25th November 2016)
OUTLINE
1. INTRODUCTION
2. RANDOM SAMPLING FOR APPROXIMATIONS
3. STATISTICAL LEARNING THEORY
‣ representativeness of a sample
‣ Rademacher averages
4. EXPERIMENTS AND RESULTS
2
BETWEENNESS CENTRALITY (BC)
▸ unweighted graph G = (V, E)
▸ n = |V|, m = |E|
3
b(w) =
1
|V | (|V | −1)
∑(u,v)∈VXV
σuv (w)
σuv
W
V
σuv (w) - number of shortest paths from u to v
passing through w U
BETWEENNESS CENTRALITY (BC)
▸ unweighted graph G = (V, E)
▸ n = |V|, m = |E|
▸ fastest exact betweenness calculation
algorithm runs in O(nm) [Brandes 2001]
▸ requires O(n+m) space
4
b(w) =
1
|V | (|V | −1)
∑(u,v)∈VXV
σuv (w)
σuv
W
V
σuv (w) - number of shortest paths from u to v
passing through w U
▸ these methods are based on random sampling to estimate
betweenness centrality with an acceptable accuracy
▸ problem definition
▸ given ε, δ ∈ (0, 1), an (ε, δ) approximation to B is a
collection such that
APPROXIMATE BC FOR LARGE NETWORKS 5
CONTRIBUTIONS OF THIS PAPER
▸ progressive sampling based BC approximation within ε
additive factor
▸ first BC approximation algorithm to estimate BC without
depending on any global property of the graph
▸ ie: RK algorithm [Riandato and Karnopoulis 2016]
depends on Vertex diameter of the graph
6
RANDOM SAMPLING TO APPROXIMATE BETWEENNESS 7
PROGRESSIVE SAMPLING 8
PROGRESSIVE SAMPLING
▸ What is a good stopping condition?
▸ guarantees that the computed approximation fulfills the
desired quality properties
▸ can be evaluated efficiently
▸ is tight (satisfied at small sample sizes)
▸ Determining sampling schedule
▸ minimize the number of iterations that are needed
before the stopping condition is satisfied
9
RECAP OF STATISTICAL LEARNING THEORY
▸ A training set S is called (w.r.t. domain Z ,
hypothesis class H , loss function l , and distribution D ) if
▸ representativeness of sample S with respect to F is
defined as the largest gap between the true error of a
function f and its empirical error
10
ε − representative
sup
h∈H
| LD (h)− LS (h)| ≤ ε
LD ( f ) = EZ~D[ f (z)] LS ( f ) =
1
m
f
i=1
m
∑ (zi )
RepD (F,S) = sup
f ∈F
(LD ( f )− LS ( f ))
given f ∈F,
REPRESENTATIVENESS OF A SAMPLE
▸ how to estimate representative of S using a single sample?
11
S =
S = sup
f ∈F
(LS1
( f )− LS2
( f ))
S = 2
m
sup
f ∈F
σi
i=1
m
∑ f (zi )
σ = (σ1,..,σm ) ∈{±1}m
RADEMACHER AVERAGE 12
‣ Rademacher complexity measure
captures this idea by considering
the expectation of the above with
respect to a random choice of σ
F°S = {( f (z1),...., f (zm )): f ∈F}
R(F°S) =
1
m
Eσ ~{±1}[sup
f ∈F
σi
i=1
m
∑ f (zi )]
σ be distributed i.i.d. according to P[i = 1] = P[i = 1] = 0.5
LD ( f )− LS ( f ) ≤ 2E ′S ~Dm R(F° ′S )+ c
2ln(2 /δ )
m
BACK TO BC
‣for each node w, is the fraction of shortest paths from u
to v going through w
13
fw (u,v)
LD ( fw ) =
1
| D |
σuv (w)
σuv(u,v)∈VXV ,u≠v
∑ = b(w)
RADEMACHER AVERAGE: HOW TO CALCULATE?
▸ calculation is not straightforward and can be time
consuming
▸ an upper bound to the Rademacher average is used in
place of
14
R(F°S) =
1
m
Eσ ~{±1}[sup
f ∈F
σi
i=1
m
∑ f (zi )]
R(F°S) ≤ mins∈!+
ω(s)
ω(s) =
1
s
ln v∈υs
e∑ xp(s2
|| v ||2
/(2m2
))
vw = ( fw (u1,v1),..., fw (um,vm ))
νs = {vw,w ∈V} (|νs |≤|V |)
R(F°S)
STOPPING CONDITION OF BC CALCULATION
▸ a tighter upper bound to maximum deviation average
calculated [Oneto 2013]
15
Δs =
ω*
1−α
+
ln(2 /δ )
2lα(1−α)
+
ln(2 /δ )
2m
Δs ≤ ε
α =
ln(2 /δ )
ln(2 /δ )+ (2lR(F°S)+ ln(2 /δ ))ln(2 /δ )
‣ when this holds collection
is returned
SAMPLING SCHEDULE
▸ initial sample size determined by
▸ next sample size ( ) is calculated assuming that , which
is and upper bound to is also an upper bound to
16
R(F°Si )
R(F°Si+1)
Si+1
DYNAMIC GRAPH BC APPROXIMATION (ABRA-D)
▸ vertex and edge insertions and deletions allowed
▸ two data structures introduced by Hayashi et al (2015)
used
▸ Hypergraph sketch: weighted hyper edge
representation of shortest paths
▸ Two-ball index: to efficiently detect the parts of the
Hypergraph sketch that need to be modified
17
EXPERIMENTAL EVALUATION
▸ performance measured using
▸ runtime
▸ sample size
▸ accuracy
▸ algorithms compared
▸ BA [Brandes 2001] - exact algorithm
▸ RK [Riondato and Kornaropoulos 2016]
18
EXPERIMENTAL RESULTS
▸ δ is is fixed to 0.1
▸ given the logarithmic dependence of the sample size on
δ, impact on the results is limited
19
REFERENCES
[1] U. Brandes. A faster algorithm for betweenness centrality. J. Math. Sociol.,
25(2):163–177, 2001. doi: 10.1080/0022250X.2001.9990249
[2] M. Riondato and E. M. Kornaropoulos. Fast approximation of betweenness
centrality through sampling. Data Mining and Knowledge Discovery, 30(2):438–
475, 2015. ISSN 1573-756X. doi: 10.1007/s10618-015-0423-0.
[3] T. Hayashi, T. Akiba, and Y. Yoshida. Fully dynamic betweenness centrality
maintenance on massive networks. Proceedings of the VLDB Endowment, 9(2),
2015
[4] L. Oneto, A. Ghio, D. Anguita, and S. Ridella. An improved analysis of the
Rademacher data-dependent bound using its self bounding property. Neural
Networks, 44:107–111, 2013.
20

Weitere ähnliche Inhalte

Was ist angesagt?

divergence of vector and divergence theorem
divergence of vector and divergence theoremdivergence of vector and divergence theorem
divergence of vector and divergence theoremAbhishekLalkiya
 
Numerical integration;Gaussian integration one point, two point and three poi...
Numerical integration;Gaussian integration one point, two point and three poi...Numerical integration;Gaussian integration one point, two point and three poi...
Numerical integration;Gaussian integration one point, two point and three poi...vaibhav tailor
 
Gaussian quadratures
Gaussian quadraturesGaussian quadratures
Gaussian quadraturesTarun Gehlot
 
Divergence Theorem by Himanshu Vaid
Divergence Theorem by Himanshu VaidDivergence Theorem by Himanshu Vaid
Divergence Theorem by Himanshu VaidHimanshu Vaid
 
Numerical integration
Numerical integrationNumerical integration
Numerical integrationTarun Gehlot
 
Line integral,Strokes and Green Theorem
Line integral,Strokes and Green TheoremLine integral,Strokes and Green Theorem
Line integral,Strokes and Green TheoremHassan Ahmed
 
Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...
Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...
Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...inventionjournals
 
Coordinate systems (and transformations) and vector calculus
Coordinate systems (and transformations) and vector calculus Coordinate systems (and transformations) and vector calculus
Coordinate systems (and transformations) and vector calculus garghanish
 
5. lec5 curl of a vector
5. lec5 curl of a vector5. lec5 curl of a vector
5. lec5 curl of a vectorshabdrang
 

Was ist angesagt? (20)

divergence of vector and divergence theorem
divergence of vector and divergence theoremdivergence of vector and divergence theorem
divergence of vector and divergence theorem
 
Kriging
KrigingKriging
Kriging
 
Numerical integration;Gaussian integration one point, two point and three poi...
Numerical integration;Gaussian integration one point, two point and three poi...Numerical integration;Gaussian integration one point, two point and three poi...
Numerical integration;Gaussian integration one point, two point and three poi...
 
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
 
Divrgence theorem with example
Divrgence theorem with exampleDivrgence theorem with example
Divrgence theorem with example
 
Gaussian quadratures
Gaussian quadraturesGaussian quadratures
Gaussian quadratures
 
Divergence Theorem by Himanshu Vaid
Divergence Theorem by Himanshu VaidDivergence Theorem by Himanshu Vaid
Divergence Theorem by Himanshu Vaid
 
Numerical integration
Numerical integrationNumerical integration
Numerical integration
 
Line integral,Strokes and Green Theorem
Line integral,Strokes and Green TheoremLine integral,Strokes and Green Theorem
Line integral,Strokes and Green Theorem
 
Regression 0410
Regression 0410Regression 0410
Regression 0410
 
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
 
Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...
Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...
Establishment of New Special Deductions from Gauss Divergence Theorem in a Ve...
 
Line integrals
Line integralsLine integrals
Line integrals
 
Multiple regression
Multiple regressionMultiple regression
Multiple regression
 
Basics1variogram
Basics1variogramBasics1variogram
Basics1variogram
 
Chapter 16 2
Chapter 16 2Chapter 16 2
Chapter 16 2
 
Chapter 16 1
Chapter 16 1Chapter 16 1
Chapter 16 1
 
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
CLIM Fall 2017 Course: Statistics for Climate Research, Statistics of Climate...
 
Coordinate systems (and transformations) and vector calculus
Coordinate systems (and transformations) and vector calculus Coordinate systems (and transformations) and vector calculus
Coordinate systems (and transformations) and vector calculus
 
5. lec5 curl of a vector
5. lec5 curl of a vector5. lec5 curl of a vector
5. lec5 curl of a vector
 

Andere mochten auch

Leveraging mobile network big data for urban planning
Leveraging mobile network big data for urban planningLeveraging mobile network big data for urban planning
Leveraging mobile network big data for urban planningKaushalya Madhawa
 
Automatic generation of event summaries using microblog streams
Automatic generation of event summaries using microblog streamsAutomatic generation of event summaries using microblog streams
Automatic generation of event summaries using microblog streamsKaushalya Madhawa
 
Opportunities in Higher Education & Career Guidance
Opportunities in Higher Education & Career GuidanceOpportunities in Higher Education & Career Guidance
Opportunities in Higher Education & Career GuidanceKaushalya Madhawa
 
Pruning convolutional neural networks for resource efficient inference
Pruning convolutional neural networks for resource efficient inferencePruning convolutional neural networks for resource efficient inference
Pruning convolutional neural networks for resource efficient inferenceKaushalya Madhawa
 
Configurando un repositorio de git hub usando netbeans ide
Configurando un repositorio de git hub usando netbeans ideConfigurando un repositorio de git hub usando netbeans ide
Configurando un repositorio de git hub usando netbeans ideferosorno
 
Understanding social connections
Understanding social connectionsUnderstanding social connections
Understanding social connectionsKaushalya Madhawa
 
An Introduction to Optimal Transport
An Introduction to Optimal TransportAn Introduction to Optimal Transport
An Introduction to Optimal TransportGabriel Peyré
 
Probabilistic Data Structures and Approximate Solutions
Probabilistic Data Structures and Approximate SolutionsProbabilistic Data Structures and Approximate Solutions
Probabilistic Data Structures and Approximate SolutionsOleksandr Pryymak
 
Neural word embedding as implicit matrix factorization の論文紹介
Neural word embedding as implicit matrix factorization の論文紹介Neural word embedding as implicit matrix factorization の論文紹介
Neural word embedding as implicit matrix factorization の論文紹介Masanao Ochi
 
Dbda勉強会~概要説明ochi20130803
Dbda勉強会~概要説明ochi20130803Dbda勉強会~概要説明ochi20130803
Dbda勉強会~概要説明ochi20130803Masanao Ochi
 
[DL輪読会]Unsupervised Learning of 3D Structure from Images
[DL輪読会]Unsupervised Learning of 3D Structure from Images[DL輪読会]Unsupervised Learning of 3D Structure from Images
[DL輪読会]Unsupervised Learning of 3D Structure from ImagesDeep Learning JP
 
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...Deep Learning JP
 
[DL輪読会]Learning to simplify fully convolutional networks for rough sketch
[DL輪読会]Learning to simplify fully convolutional networks for rough sketch[DL輪読会]Learning to simplify fully convolutional networks for rough sketch
[DL輪読会]Learning to simplify fully convolutional networks for rough sketchDeep Learning JP
 
A Gentle Introduction to Locality Sensitive Hashing with Apache Spark
A Gentle Introduction to Locality Sensitive Hashing with Apache SparkA Gentle Introduction to Locality Sensitive Hashing with Apache Spark
A Gentle Introduction to Locality Sensitive Hashing with Apache SparkFrançois Garillot
 
[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...
[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...
[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...Deep Learning JP
 
[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODER
[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODER[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODER
[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODERDeep Learning JP
 
[DL輪読会]Let there be color
[DL輪読会]Let there be color[DL輪読会]Let there be color
[DL輪読会]Let there be colorDeep Learning JP
 
[DL輪読会]Image-to-Image Translation with Conditional Adversarial Networks
[DL輪読会]Image-to-Image Translation with Conditional Adversarial Networks[DL輪読会]Image-to-Image Translation with Conditional Adversarial Networks
[DL輪読会]Image-to-Image Translation with Conditional Adversarial NetworksDeep Learning JP
 
[DL輪読会]Learning convolutional neural networks for graphs
[DL輪読会]Learning convolutional neural networks for graphs[DL輪読会]Learning convolutional neural networks for graphs
[DL輪読会]Learning convolutional neural networks for graphsDeep Learning JP
 

Andere mochten auch (20)

Leveraging mobile network big data for urban planning
Leveraging mobile network big data for urban planningLeveraging mobile network big data for urban planning
Leveraging mobile network big data for urban planning
 
Automatic generation of event summaries using microblog streams
Automatic generation of event summaries using microblog streamsAutomatic generation of event summaries using microblog streams
Automatic generation of event summaries using microblog streams
 
Opportunities in Higher Education & Career Guidance
Opportunities in Higher Education & Career GuidanceOpportunities in Higher Education & Career Guidance
Opportunities in Higher Education & Career Guidance
 
Pruning convolutional neural networks for resource efficient inference
Pruning convolutional neural networks for resource efficient inferencePruning convolutional neural networks for resource efficient inference
Pruning convolutional neural networks for resource efficient inference
 
Configurando un repositorio de git hub usando netbeans ide
Configurando un repositorio de git hub usando netbeans ideConfigurando un repositorio de git hub usando netbeans ide
Configurando un repositorio de git hub usando netbeans ide
 
Understanding social connections
Understanding social connectionsUnderstanding social connections
Understanding social connections
 
An Introduction to Optimal Transport
An Introduction to Optimal TransportAn Introduction to Optimal Transport
An Introduction to Optimal Transport
 
Probabilistic Data Structures and Approximate Solutions
Probabilistic Data Structures and Approximate SolutionsProbabilistic Data Structures and Approximate Solutions
Probabilistic Data Structures and Approximate Solutions
 
AISTAT2016 SNFS
AISTAT2016 SNFSAISTAT2016 SNFS
AISTAT2016 SNFS
 
Neural word embedding as implicit matrix factorization の論文紹介
Neural word embedding as implicit matrix factorization の論文紹介Neural word embedding as implicit matrix factorization の論文紹介
Neural word embedding as implicit matrix factorization の論文紹介
 
Dbda勉強会~概要説明ochi20130803
Dbda勉強会~概要説明ochi20130803Dbda勉強会~概要説明ochi20130803
Dbda勉強会~概要説明ochi20130803
 
[DL輪読会]Unsupervised Learning of 3D Structure from Images
[DL輪読会]Unsupervised Learning of 3D Structure from Images[DL輪読会]Unsupervised Learning of 3D Structure from Images
[DL輪読会]Unsupervised Learning of 3D Structure from Images
 
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...
[Dl輪読会]bridging the gaps between residual learning, recurrent neural networks...
 
[DL輪読会]Learning to simplify fully convolutional networks for rough sketch
[DL輪読会]Learning to simplify fully convolutional networks for rough sketch[DL輪読会]Learning to simplify fully convolutional networks for rough sketch
[DL輪読会]Learning to simplify fully convolutional networks for rough sketch
 
A Gentle Introduction to Locality Sensitive Hashing with Apache Spark
A Gentle Introduction to Locality Sensitive Hashing with Apache SparkA Gentle Introduction to Locality Sensitive Hashing with Apache Spark
A Gentle Introduction to Locality Sensitive Hashing with Apache Spark
 
[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...
[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...
[DL輪読会]Combining Fully Convolutional and Recurrent Neural Networks for 3D Bio...
 
[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODER
[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODER[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODER
[DL輪読会]TREE-STRUCTURED VARIATIONAL AUTOENCODER
 
[DL輪読会]Let there be color
[DL輪読会]Let there be color[DL輪読会]Let there be color
[DL輪読会]Let there be color
 
[DL輪読会]Image-to-Image Translation with Conditional Adversarial Networks
[DL輪読会]Image-to-Image Translation with Conditional Adversarial Networks[DL輪読会]Image-to-Image Translation with Conditional Adversarial Networks
[DL輪読会]Image-to-Image Translation with Conditional Adversarial Networks
 
[DL輪読会]Learning convolutional neural networks for graphs
[DL輪読会]Learning convolutional neural networks for graphs[DL輪読会]Learning convolutional neural networks for graphs
[DL輪読会]Learning convolutional neural networks for graphs
 

Ähnlich wie ABRA: Approximating Betweenness Centrality in Static and Dynamic Graphs with Rademacher Averages

Fixed points theorem on a pair of random generalized non linear contractions
Fixed points theorem on a pair of random generalized non linear contractionsFixed points theorem on a pair of random generalized non linear contractions
Fixed points theorem on a pair of random generalized non linear contractionsAlexander Decker
 
Litvinenko_RWTH_UQ_Seminar_talk.pdf
Litvinenko_RWTH_UQ_Seminar_talk.pdfLitvinenko_RWTH_UQ_Seminar_talk.pdf
Litvinenko_RWTH_UQ_Seminar_talk.pdfAlexander Litvinenko
 
Rademacher Averages: Theory and Practice
Rademacher Averages: Theory and PracticeRademacher Averages: Theory and Practice
Rademacher Averages: Theory and PracticeTwo Sigma
 
A Szemeredi-type theorem for subsets of the unit cube
A Szemeredi-type theorem for subsets of the unit cubeA Szemeredi-type theorem for subsets of the unit cube
A Szemeredi-type theorem for subsets of the unit cubeVjekoslavKovac1
 
ABC with Wasserstein distances
ABC with Wasserstein distancesABC with Wasserstein distances
ABC with Wasserstein distancesChristian Robert
 
QMC Error SAMSI Tutorial Aug 2017
QMC Error SAMSI Tutorial Aug 2017QMC Error SAMSI Tutorial Aug 2017
QMC Error SAMSI Tutorial Aug 2017Fred J. Hickernell
 
On maximal and variational Fourier restriction
On maximal and variational Fourier restrictionOn maximal and variational Fourier restriction
On maximal and variational Fourier restrictionVjekoslavKovac1
 
Single source shortes path in dag
Single source shortes path in dagSingle source shortes path in dag
Single source shortes path in dagKiran K
 
Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...Valentin De Bortoli
 
Patch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective DivergencesPatch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective DivergencesFrank Nielsen
 
On learning statistical mixtures maximizing the complete likelihood
On learning statistical mixtures maximizing the complete likelihoodOn learning statistical mixtures maximizing the complete likelihood
On learning statistical mixtures maximizing the complete likelihoodFrank Nielsen
 
Darmon Points: an Overview
Darmon Points: an OverviewDarmon Points: an Overview
Darmon Points: an Overviewmmasdeu
 
l1-Embeddings and Algorithmic Applications
l1-Embeddings and Algorithmic Applicationsl1-Embeddings and Algorithmic Applications
l1-Embeddings and Algorithmic ApplicationsGrigory Yaroslavtsev
 
線形回帰モデル
線形回帰モデル線形回帰モデル
線形回帰モデル貴之 八木
 

Ähnlich wie ABRA: Approximating Betweenness Centrality in Static and Dynamic Graphs with Rademacher Averages (20)

MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models...
MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models...MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models...
MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models...
 
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
 
Fixed points theorem on a pair of random generalized non linear contractions
Fixed points theorem on a pair of random generalized non linear contractionsFixed points theorem on a pair of random generalized non linear contractions
Fixed points theorem on a pair of random generalized non linear contractions
 
QMC: Operator Splitting Workshop, Proximal Algorithms in Probability Spaces -...
QMC: Operator Splitting Workshop, Proximal Algorithms in Probability Spaces -...QMC: Operator Splitting Workshop, Proximal Algorithms in Probability Spaces -...
QMC: Operator Splitting Workshop, Proximal Algorithms in Probability Spaces -...
 
Litvinenko_RWTH_UQ_Seminar_talk.pdf
Litvinenko_RWTH_UQ_Seminar_talk.pdfLitvinenko_RWTH_UQ_Seminar_talk.pdf
Litvinenko_RWTH_UQ_Seminar_talk.pdf
 
PCA on graph/network
PCA on graph/networkPCA on graph/network
PCA on graph/network
 
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
 
Rademacher Averages: Theory and Practice
Rademacher Averages: Theory and PracticeRademacher Averages: Theory and Practice
Rademacher Averages: Theory and Practice
 
A Szemeredi-type theorem for subsets of the unit cube
A Szemeredi-type theorem for subsets of the unit cubeA Szemeredi-type theorem for subsets of the unit cube
A Szemeredi-type theorem for subsets of the unit cube
 
ABC with Wasserstein distances
ABC with Wasserstein distancesABC with Wasserstein distances
ABC with Wasserstein distances
 
QMC Error SAMSI Tutorial Aug 2017
QMC Error SAMSI Tutorial Aug 2017QMC Error SAMSI Tutorial Aug 2017
QMC Error SAMSI Tutorial Aug 2017
 
On maximal and variational Fourier restriction
On maximal and variational Fourier restrictionOn maximal and variational Fourier restriction
On maximal and variational Fourier restriction
 
Single source shortes path in dag
Single source shortes path in dagSingle source shortes path in dag
Single source shortes path in dag
 
multiple intrigral lit
multiple intrigral litmultiple intrigral lit
multiple intrigral lit
 
Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...
 
Patch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective DivergencesPatch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective Divergences
 
On learning statistical mixtures maximizing the complete likelihood
On learning statistical mixtures maximizing the complete likelihoodOn learning statistical mixtures maximizing the complete likelihood
On learning statistical mixtures maximizing the complete likelihood
 
Darmon Points: an Overview
Darmon Points: an OverviewDarmon Points: an Overview
Darmon Points: an Overview
 
l1-Embeddings and Algorithmic Applications
l1-Embeddings and Algorithmic Applicationsl1-Embeddings and Algorithmic Applications
l1-Embeddings and Algorithmic Applications
 
線形回帰モデル
線形回帰モデル線形回帰モデル
線形回帰モデル
 

Kürzlich hochgeladen

04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationshipsccctableauusergroup
 
B2 Creative Industry Response Evaluation.docx
B2 Creative Industry Response Evaluation.docxB2 Creative Industry Response Evaluation.docx
B2 Creative Industry Response Evaluation.docxStephen266013
 
Smarteg dropshipping via API with DroFx.pptx
Smarteg dropshipping via API with DroFx.pptxSmarteg dropshipping via API with DroFx.pptx
Smarteg dropshipping via API with DroFx.pptxolyaivanovalion
 
BigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxBigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxolyaivanovalion
 
定制英国白金汉大学毕业证(UCB毕业证书) 成绩单原版一比一
定制英国白金汉大学毕业证(UCB毕业证书)																			成绩单原版一比一定制英国白金汉大学毕业证(UCB毕业证书)																			成绩单原版一比一
定制英国白金汉大学毕业证(UCB毕业证书) 成绩单原版一比一ffjhghh
 
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptxBPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptxMohammedJunaid861692
 
VidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptxVidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptxolyaivanovalion
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...apidays
 
Low Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service Bhilai
Low Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service BhilaiLow Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service Bhilai
Low Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service BhilaiSuhani Kapoor
 
Industrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfIndustrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfLars Albertsson
 
Generative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusGenerative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusTimothy Spann
 
Invezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signalsInvezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signalsInvezz1
 
April 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's AnalysisApril 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's Analysismanisha194592
 
Log Analysis using OSSEC sasoasasasas.pptx
Log Analysis using OSSEC sasoasasasas.pptxLog Analysis using OSSEC sasoasasasas.pptx
Log Analysis using OSSEC sasoasasasas.pptxJohnnyPlasten
 
Data-Analysis for Chicago Crime Data 2023
Data-Analysis for Chicago Crime Data  2023Data-Analysis for Chicago Crime Data  2023
Data-Analysis for Chicago Crime Data 2023ymrp368
 
Halmar dropshipping via API with DroFx
Halmar  dropshipping  via API with DroFxHalmar  dropshipping  via API with DroFx
Halmar dropshipping via API with DroFxolyaivanovalion
 
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Callshivangimorya083
 
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAl Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAroojKhan71
 

Kürzlich hochgeladen (20)

04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships
 
B2 Creative Industry Response Evaluation.docx
B2 Creative Industry Response Evaluation.docxB2 Creative Industry Response Evaluation.docx
B2 Creative Industry Response Evaluation.docx
 
VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...
VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...
VIP Call Girls Service Charbagh { Lucknow Call Girls Service 9548273370 } Boo...
 
Smarteg dropshipping via API with DroFx.pptx
Smarteg dropshipping via API with DroFx.pptxSmarteg dropshipping via API with DroFx.pptx
Smarteg dropshipping via API with DroFx.pptx
 
Sampling (random) method and Non random.ppt
Sampling (random) method and Non random.pptSampling (random) method and Non random.ppt
Sampling (random) method and Non random.ppt
 
BigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxBigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptx
 
定制英国白金汉大学毕业证(UCB毕业证书) 成绩单原版一比一
定制英国白金汉大学毕业证(UCB毕业证书)																			成绩单原版一比一定制英国白金汉大学毕业证(UCB毕业证书)																			成绩单原版一比一
定制英国白金汉大学毕业证(UCB毕业证书) 成绩单原版一比一
 
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptxBPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
 
VidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptxVidaXL dropshipping via API with DroFx.pptx
VidaXL dropshipping via API with DroFx.pptx
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
 
Low Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service Bhilai
Low Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service BhilaiLow Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service Bhilai
Low Rate Call Girls Bhilai Anika 8250192130 Independent Escort Service Bhilai
 
Industrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfIndustrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdf
 
Generative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusGenerative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and Milvus
 
Invezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signalsInvezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signals
 
April 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's AnalysisApril 2024 - Crypto Market Report's Analysis
April 2024 - Crypto Market Report's Analysis
 
Log Analysis using OSSEC sasoasasasas.pptx
Log Analysis using OSSEC sasoasasasas.pptxLog Analysis using OSSEC sasoasasasas.pptx
Log Analysis using OSSEC sasoasasasas.pptx
 
Data-Analysis for Chicago Crime Data 2023
Data-Analysis for Chicago Crime Data  2023Data-Analysis for Chicago Crime Data  2023
Data-Analysis for Chicago Crime Data 2023
 
Halmar dropshipping via API with DroFx
Halmar  dropshipping  via API with DroFxHalmar  dropshipping  via API with DroFx
Halmar dropshipping via API with DroFx
 
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
 
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAl Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
 

ABRA: Approximating Betweenness Centrality in Static and Dynamic Graphs with Rademacher Averages

  • 1. ABRA: APPROXIMATING BETWEENNESS CENTRALITY IN STATIC AND DYNAMIC GRAPHS WITH RADEMACHER AVERAGES Matteo Riondata and Eli Upfal 22nd ACM SIGKDD Conference, August 2016 1 Murata Lab - Paper reading seminar Presented by: Kaushalya Madhawa (25th November 2016)
  • 2. OUTLINE 1. INTRODUCTION 2. RANDOM SAMPLING FOR APPROXIMATIONS 3. STATISTICAL LEARNING THEORY ‣ representativeness of a sample ‣ Rademacher averages 4. EXPERIMENTS AND RESULTS 2
  • 3. BETWEENNESS CENTRALITY (BC) ▸ unweighted graph G = (V, E) ▸ n = |V|, m = |E| 3 b(w) = 1 |V | (|V | −1) ∑(u,v)∈VXV σuv (w) σuv W V σuv (w) - number of shortest paths from u to v passing through w U
  • 4. BETWEENNESS CENTRALITY (BC) ▸ unweighted graph G = (V, E) ▸ n = |V|, m = |E| ▸ fastest exact betweenness calculation algorithm runs in O(nm) [Brandes 2001] ▸ requires O(n+m) space 4 b(w) = 1 |V | (|V | −1) ∑(u,v)∈VXV σuv (w) σuv W V σuv (w) - number of shortest paths from u to v passing through w U
  • 5. ▸ these methods are based on random sampling to estimate betweenness centrality with an acceptable accuracy ▸ problem definition ▸ given ε, δ ∈ (0, 1), an (ε, δ) approximation to B is a collection such that APPROXIMATE BC FOR LARGE NETWORKS 5
  • 6. CONTRIBUTIONS OF THIS PAPER ▸ progressive sampling based BC approximation within ε additive factor ▸ first BC approximation algorithm to estimate BC without depending on any global property of the graph ▸ ie: RK algorithm [Riandato and Karnopoulis 2016] depends on Vertex diameter of the graph 6
  • 7. RANDOM SAMPLING TO APPROXIMATE BETWEENNESS 7
  • 9. PROGRESSIVE SAMPLING ▸ What is a good stopping condition? ▸ guarantees that the computed approximation fulfills the desired quality properties ▸ can be evaluated efficiently ▸ is tight (satisfied at small sample sizes) ▸ Determining sampling schedule ▸ minimize the number of iterations that are needed before the stopping condition is satisfied 9
  • 10. RECAP OF STATISTICAL LEARNING THEORY ▸ A training set S is called (w.r.t. domain Z , hypothesis class H , loss function l , and distribution D ) if ▸ representativeness of sample S with respect to F is defined as the largest gap between the true error of a function f and its empirical error 10 ε − representative sup h∈H | LD (h)− LS (h)| ≤ ε LD ( f ) = EZ~D[ f (z)] LS ( f ) = 1 m f i=1 m ∑ (zi ) RepD (F,S) = sup f ∈F (LD ( f )− LS ( f )) given f ∈F,
  • 11. REPRESENTATIVENESS OF A SAMPLE ▸ how to estimate representative of S using a single sample? 11 S = S = sup f ∈F (LS1 ( f )− LS2 ( f )) S = 2 m sup f ∈F σi i=1 m ∑ f (zi ) σ = (σ1,..,σm ) ∈{±1}m
  • 12. RADEMACHER AVERAGE 12 ‣ Rademacher complexity measure captures this idea by considering the expectation of the above with respect to a random choice of σ F°S = {( f (z1),...., f (zm )): f ∈F} R(F°S) = 1 m Eσ ~{±1}[sup f ∈F σi i=1 m ∑ f (zi )] σ be distributed i.i.d. according to P[i = 1] = P[i = 1] = 0.5 LD ( f )− LS ( f ) ≤ 2E ′S ~Dm R(F° ′S )+ c 2ln(2 /δ ) m
  • 13. BACK TO BC ‣for each node w, is the fraction of shortest paths from u to v going through w 13 fw (u,v) LD ( fw ) = 1 | D | σuv (w) σuv(u,v)∈VXV ,u≠v ∑ = b(w)
  • 14. RADEMACHER AVERAGE: HOW TO CALCULATE? ▸ calculation is not straightforward and can be time consuming ▸ an upper bound to the Rademacher average is used in place of 14 R(F°S) = 1 m Eσ ~{±1}[sup f ∈F σi i=1 m ∑ f (zi )] R(F°S) ≤ mins∈!+ ω(s) ω(s) = 1 s ln v∈υs e∑ xp(s2 || v ||2 /(2m2 )) vw = ( fw (u1,v1),..., fw (um,vm )) νs = {vw,w ∈V} (|νs |≤|V |) R(F°S)
  • 15. STOPPING CONDITION OF BC CALCULATION ▸ a tighter upper bound to maximum deviation average calculated [Oneto 2013] 15 Δs = ω* 1−α + ln(2 /δ ) 2lα(1−α) + ln(2 /δ ) 2m Δs ≤ ε α = ln(2 /δ ) ln(2 /δ )+ (2lR(F°S)+ ln(2 /δ ))ln(2 /δ ) ‣ when this holds collection is returned
  • 16. SAMPLING SCHEDULE ▸ initial sample size determined by ▸ next sample size ( ) is calculated assuming that , which is and upper bound to is also an upper bound to 16 R(F°Si ) R(F°Si+1) Si+1
  • 17. DYNAMIC GRAPH BC APPROXIMATION (ABRA-D) ▸ vertex and edge insertions and deletions allowed ▸ two data structures introduced by Hayashi et al (2015) used ▸ Hypergraph sketch: weighted hyper edge representation of shortest paths ▸ Two-ball index: to efficiently detect the parts of the Hypergraph sketch that need to be modified 17
  • 18. EXPERIMENTAL EVALUATION ▸ performance measured using ▸ runtime ▸ sample size ▸ accuracy ▸ algorithms compared ▸ BA [Brandes 2001] - exact algorithm ▸ RK [Riondato and Kornaropoulos 2016] 18
  • 19. EXPERIMENTAL RESULTS ▸ δ is is fixed to 0.1 ▸ given the logarithmic dependence of the sample size on δ, impact on the results is limited 19
  • 20. REFERENCES [1] U. Brandes. A faster algorithm for betweenness centrality. J. Math. Sociol., 25(2):163–177, 2001. doi: 10.1080/0022250X.2001.9990249 [2] M. Riondato and E. M. Kornaropoulos. Fast approximation of betweenness centrality through sampling. Data Mining and Knowledge Discovery, 30(2):438– 475, 2015. ISSN 1573-756X. doi: 10.1007/s10618-015-0423-0. [3] T. Hayashi, T. Akiba, and Y. Yoshida. Fully dynamic betweenness centrality maintenance on massive networks. Proceedings of the VLDB Endowment, 9(2), 2015 [4] L. Oneto, A. Ghio, D. Anguita, and S. Ridella. An improved analysis of the Rademacher data-dependent bound using its self bounding property. Neural Networks, 44:107–111, 2013. 20