SlideShare a Scribd company logo
1 of 41
Download to read offline
Graph Kernel
CS15MTECH11011
THESIS(STAGE I)
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
Kernels
 A kernel is a function of two objects that quantifies their similarity.
Any kernel methods solution comprises two parts: a module that performs the mapping into the
embedding or feature space and a learning algorithm designed to discover linear patterns in that
space.
K(X,X’)=(Φ(X),Φ(X’))
Where Φ is the feature space transformation.
Property: Symmetric Positive-Definite
C’KC≥0;
Graph Kernel
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
Graph Kernel
Graph kernels are used to find similarity between two graphs.
It exploits graph topology but, restricts itself to comparing substructure of graph that are
computable in polynomial time.
Property: Symmetric positive-Definite.
Many Graph Kernels have been defined, we will focus on mainly following:
1. Shortest Path Kernel
2. Graphlet Kernel
3. Weisfeiler-Lehman Kernel.
GRAPH KERNEL
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
Shortest Path Kernel
Task: Find the shortest path of all lengths among the graphs.
Working:
1. Floyd Transformation of graph.
2. Apply shortest path kernel over that transformed graph.
Complexity: O(n4)
GRAPH KERNEL
SP Kernel(Cont.)
Floyd Transformation:
 Contains same number of nodes as input graph.
 Unlike, there exist an edge between nodes in transformed graph if there exist a walk
between the nodes in input graph.
 Edge are marked as shortest distance
between nodes
 Complexity O(n3) Floyd-Warshall algorithm
GRAPH KERNEL
SP Kernel(Cont.)
GRAPH KERNEL
Kernel: Let S1(E1,V1), S2(E2,V2) be two graphs
SP Kernel (Cont.)
GRAPH KERNEL
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
Graphlet Kernel
Task: Count number of graphlet because isomorphic graphs have same graphlet distribution.
Count the number of graphlets of specific size, for linear computation.
There are possible graphlet of size k.
Let G and G’ be 2 graphs and S and S’ be their respective subgraphs.
Complexity: O(ndk-1)
 d=maximum degree of node
 k= size of graphlet
GRAPH KERNEL
Graphlet Kernel
 computing 3-graphlet kernel for connected component (triangle, V shapes).
GRAPH KERNEL
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
Weisfeiler-Lehman Kernel
WL Algorithm:
GRAPH KERNEL
WL Algorithm :iteration 1
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
GRAPH KERNEL
WL Algorithm :iteration 1
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
GRAPH KERNEL
WL Algorithm :iteration 1
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
GRAPH KERNEL
WL Algorithm :iteration 1
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
3. Relabeling
◦ O(n)
GRAPH KERNEL
WL Algorithm :iteration 1
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
3. Relabeling
◦ O(n)
4.Are the labels of G and G’ identical?
Yes, continue.
GRAPH KERNEL
WL Algorithm :iteration 2
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
GRAPH KERNEL
WL Algorithm :iteration 2
Each Iteration of WL test comprises of following steps:
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
GRAPH KERNEL
WL Algorithm :iteration 2
Each Iteration of WL test comprises of following steps:
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
GRAPH KERNEL
WL Algorithm :iteration 2
Each Iteration of WL test comprises of following steps:
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
GRAPH KERNEL
WL Algorithm :iteration 2
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
3. Relabeling
◦ O(n)
GRAPH KERNEL
WL Algorithm :iteration 2
Each Iteration of WL test comprises of following steps:-
1.Multiset label determination and sorting
◦ O(m) via Bucket Sort
2.label compression
◦ O(m) via Radix Sort
3. Relabeling
◦ O(n)
4.Are the labels of G and G’ identical?
NO, output YES.
5.complexity O(hm) for h iteration
GRAPH KERNEL
WL Kernel
Feature{2 3 4 5 6 7 8 9 10}
Φ(G1)={3 2 1 0 1 2 2 0 1}
Φ(G2)={3 2 1 2 1 0 0 2 1 }
K{G1,G2}=16
GRAPH KERNEL
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
EXPERIMENT
1. INPUT: We have generated 3 different kind of input graphs.
a)Edge Removed Graphs: We took a complete graph and removed 10 edges at a time
from that graph, iteratively generate new graph.
b)Degree Reduced Graph: We took a complete graph and reduced the degree of each
node by 1,iteratively generate new graph.
c)Node Removed Graph: We took a complete graph and removed 1 node at a time from
that graph to generate new graph iteratively.
GRAPH KERNEL
EXPERIMENT
NORMALIZATION:
GRAPH KERNEL
Graphlet Kernel norm.
GRAPH KERNEL
WL Kernel norm.
GRAPH KERNEL
Shortest Path Kernel norm.
GRAPH KERNEL
SVM
Graph Label Assignment:10 belong to each class, iteratively.
10 Fold SVM
Function Used: ‘rbf’
Gamma=0.00001
Accuracy: mean Accuracy output from all 10 folds.
GRAPH KERNEL
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
GRAPH KERNEL
GRAPH KERNEL
GRAPH KERNEL
Outline
oKernel
oGraph Kernel
oShortest Path Kernel
oGraphlet Kernel
oWL Kernel
oExperimentation
oResults
oConclusion
Conclusion
Accuracy:
Graphlet Kernel > WL Kernel > Shortest Path Kernel
I. Graphlet Kernel Exploits the topology very effectively by considering graphlet distribution.
II. Shortest Path Kernel shows least accuracy because it computes the number of shortest path
while not considering the topology involved in graphs.
III. WL Kernel shows moderate performance on our dataset.
IV. As we increase number of nodes the accuracy decreases because number of false
classification increases.
GRAPH KERNEL
Conclusion (cont.)
Time
Graphlet Kernel > Shortest Path Kernel > WL Kernel
I. Graphlet Kernel has highest computation time O(ndk-1),since our dataset is pretty dense.
II. Shortest Path Kernel O(n4).
III. WL Kernel O(Nhm).
N=number of Graphs.
h=number of iterations.
m=number of distinct labels.
GRAPH KERNEL

More Related Content

What's hot

Deep Semi-supervised Learning methods
Deep Semi-supervised Learning methodsDeep Semi-supervised Learning methods
Deep Semi-supervised Learning methodsPrincy Joy
 
Unsupervised learning clustering
Unsupervised learning clusteringUnsupervised learning clustering
Unsupervised learning clusteringArshad Farhad
 
Variational Autoencoder
Variational AutoencoderVariational Autoencoder
Variational AutoencoderMark Chang
 
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks Christopher Morris
 
07 regularization
07 regularization07 regularization
07 regularizationRonald Teo
 
Hyperparameter Optimization for Machine Learning
Hyperparameter Optimization for Machine LearningHyperparameter Optimization for Machine Learning
Hyperparameter Optimization for Machine LearningFrancesco Casalegno
 
Batch normalization presentation
Batch normalization presentationBatch normalization presentation
Batch normalization presentationOwin Will
 
Reinforcement Learning Q-Learning
Reinforcement Learning   Q-Learning Reinforcement Learning   Q-Learning
Reinforcement Learning Q-Learning Melaku Eneayehu
 
Machine learning with scikitlearn
Machine learning with scikitlearnMachine learning with scikitlearn
Machine learning with scikitlearnPratap Dangeti
 
Instance based learning
Instance based learningInstance based learning
Instance based learningswapnac12
 
Representation learning on graphs
Representation learning on graphsRepresentation learning on graphs
Representation learning on graphsDeakin University
 
Generative Adversarial Networks
Generative Adversarial NetworksGenerative Adversarial Networks
Generative Adversarial NetworksMustafa Yagmur
 

What's hot (20)

Deep Semi-supervised Learning methods
Deep Semi-supervised Learning methodsDeep Semi-supervised Learning methods
Deep Semi-supervised Learning methods
 
Gnn overview
Gnn overviewGnn overview
Gnn overview
 
K mean-clustering
K mean-clusteringK mean-clustering
K mean-clustering
 
Unsupervised learning clustering
Unsupervised learning clusteringUnsupervised learning clustering
Unsupervised learning clustering
 
Variational Autoencoder
Variational AutoencoderVariational Autoencoder
Variational Autoencoder
 
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks
 
Clustering
ClusteringClustering
Clustering
 
07 regularization
07 regularization07 regularization
07 regularization
 
Multi Layer Network
Multi Layer NetworkMulti Layer Network
Multi Layer Network
 
Hyperparameter Optimization for Machine Learning
Hyperparameter Optimization for Machine LearningHyperparameter Optimization for Machine Learning
Hyperparameter Optimization for Machine Learning
 
Batch normalization presentation
Batch normalization presentationBatch normalization presentation
Batch normalization presentation
 
Reinforcement Learning Q-Learning
Reinforcement Learning   Q-Learning Reinforcement Learning   Q-Learning
Reinforcement Learning Q-Learning
 
Machine learning with scikitlearn
Machine learning with scikitlearnMachine learning with scikitlearn
Machine learning with scikitlearn
 
Instance based learning
Instance based learningInstance based learning
Instance based learning
 
Representation learning on graphs
Representation learning on graphsRepresentation learning on graphs
Representation learning on graphs
 
Recurrent Neural Networks
Recurrent Neural NetworksRecurrent Neural Networks
Recurrent Neural Networks
 
Density based clustering
Density based clusteringDensity based clustering
Density based clustering
 
Generative Adversarial Networks
Generative Adversarial NetworksGenerative Adversarial Networks
Generative Adversarial Networks
 
Transfer Learning
Transfer LearningTransfer Learning
Transfer Learning
 
Swin transformer
Swin transformerSwin transformer
Swin transformer
 

Viewers also liked

Metodo Monte Carlo -Wang Landau
Metodo Monte Carlo -Wang LandauMetodo Monte Carlo -Wang Landau
Metodo Monte Carlo -Wang Landauangely alcendra
 
Webinar: Moving to Office 365? What You Need to Know!
Webinar: Moving to Office 365? What You Need to Know!Webinar: Moving to Office 365? What You Need to Know!
Webinar: Moving to Office 365? What You Need to Know!Flexera
 
Lei rerestrutura cms finalizada
Lei rerestrutura cms finalizadaLei rerestrutura cms finalizada
Lei rerestrutura cms finalizadaclaudio silva
 
Public cloud on System z
Public cloud on System zPublic cloud on System z
Public cloud on System zAndy Avramenko
 
Regimento interno do conselho municipal de saúde de jequié
Regimento interno do conselho municipal de saúde de jequié Regimento interno do conselho municipal de saúde de jequié
Regimento interno do conselho municipal de saúde de jequié claudio silva
 
Welcome To The Collaborative economy
Welcome To The Collaborative economyWelcome To The Collaborative economy
Welcome To The Collaborative economyAhmed Abaza
 
The Biochemica Genesis_vol 12
The Biochemica Genesis_vol 12The Biochemica Genesis_vol 12
The Biochemica Genesis_vol 12Mohit Singh Rana
 
ONE Championship - Introduction
ONE Championship - IntroductionONE Championship - Introduction
ONE Championship - IntroductionYu Hong Seah
 
Partes de un ordenador Rosi
Partes de un ordenador RosiPartes de un ordenador Rosi
Partes de un ordenador Rosinc_hdez
 
La bella durmiente
La bella durmiente La bella durmiente
La bella durmiente JRambayR
 

Viewers also liked (15)

Metodo Monte Carlo -Wang Landau
Metodo Monte Carlo -Wang LandauMetodo Monte Carlo -Wang Landau
Metodo Monte Carlo -Wang Landau
 
Webinar: Moving to Office 365? What You Need to Know!
Webinar: Moving to Office 365? What You Need to Know!Webinar: Moving to Office 365? What You Need to Know!
Webinar: Moving to Office 365? What You Need to Know!
 
Broadcast Release
Broadcast ReleaseBroadcast Release
Broadcast Release
 
Estequiometria
EstequiometriaEstequiometria
Estequiometria
 
Lei rerestrutura cms finalizada
Lei rerestrutura cms finalizadaLei rerestrutura cms finalizada
Lei rerestrutura cms finalizada
 
Public cloud on System z
Public cloud on System zPublic cloud on System z
Public cloud on System z
 
Regimento interno do conselho municipal de saúde de jequié
Regimento interno do conselho municipal de saúde de jequié Regimento interno do conselho municipal de saúde de jequié
Regimento interno do conselho municipal de saúde de jequié
 
Welcome To The Collaborative economy
Welcome To The Collaborative economyWelcome To The Collaborative economy
Welcome To The Collaborative economy
 
Azhar
AzharAzhar
Azhar
 
The Biochemica Genesis_vol 12
The Biochemica Genesis_vol 12The Biochemica Genesis_vol 12
The Biochemica Genesis_vol 12
 
ONE Championship - Introduction
ONE Championship - IntroductionONE Championship - Introduction
ONE Championship - Introduction
 
Partes de un ordenador Rosi
Partes de un ordenador RosiPartes de un ordenador Rosi
Partes de un ordenador Rosi
 
La bella durmiente
La bella durmiente La bella durmiente
La bella durmiente
 
APE Bootcamp 2015
APE Bootcamp 2015APE Bootcamp 2015
APE Bootcamp 2015
 
Portfolio
PortfolioPortfolio
Portfolio
 

Similar to Graph Kernelpdf

Monadic genetic kernels in Scala
Monadic genetic kernels in ScalaMonadic genetic kernels in Scala
Monadic genetic kernels in ScalaPatrick Nicolas
 
Learning Convolutional Neural Networks for Graphs
Learning Convolutional Neural Networks for GraphsLearning Convolutional Neural Networks for Graphs
Learning Convolutional Neural Networks for Graphspione30
 
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...ssuser2624f71
 
Anomaly detection using deep one class classifier
Anomaly detection using deep one class classifierAnomaly detection using deep one class classifier
Anomaly detection using deep one class classifier홍배 김
 
The world of loss function
The world of loss functionThe world of loss function
The world of loss function홍배 김
 
SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15Hao Zhuang
 
Learning LWF Chain Graphs: A Markov Blanket Discovery Approach
Learning LWF Chain Graphs: A Markov Blanket Discovery ApproachLearning LWF Chain Graphs: A Markov Blanket Discovery Approach
Learning LWF Chain Graphs: A Markov Blanket Discovery ApproachPooyan Jamshidi
 
Efficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketchingEfficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketchingHsing-chuan Hsieh
 
TAME: Trainable Attention Mechanism for Explanations
TAME: Trainable Attention Mechanism for ExplanationsTAME: Trainable Attention Mechanism for Explanations
TAME: Trainable Attention Mechanism for ExplanationsVasileiosMezaris
 
Pr045 deep lab_semantic_segmentation
Pr045 deep lab_semantic_segmentationPr045 deep lab_semantic_segmentation
Pr045 deep lab_semantic_segmentationTaeoh Kim
 
Barker_SIAMCSE15
Barker_SIAMCSE15Barker_SIAMCSE15
Barker_SIAMCSE15Karen Pao
 
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHMADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHMWireilla
 
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHMADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHMijfls
 
Inside LoLA - Experiences from building a state space tool for place transiti...
Inside LoLA - Experiences from building a state space tool for place transiti...Inside LoLA - Experiences from building a state space tool for place transiti...
Inside LoLA - Experiences from building a state space tool for place transiti...Universität Rostock
 
GKEL_IGARSS_2011.ppt
GKEL_IGARSS_2011.pptGKEL_IGARSS_2011.ppt
GKEL_IGARSS_2011.pptgrssieee
 
VJAI Paper Reading#3-KDD2019-ClusterGCN
VJAI Paper Reading#3-KDD2019-ClusterGCNVJAI Paper Reading#3-KDD2019-ClusterGCN
VJAI Paper Reading#3-KDD2019-ClusterGCNDat Nguyen
 
Incremental and parallel computation of structural graph summaries for evolvi...
Incremental and parallel computation of structural graph summaries for evolvi...Incremental and parallel computation of structural graph summaries for evolvi...
Incremental and parallel computation of structural graph summaries for evolvi...Till Blume
 

Similar to Graph Kernelpdf (20)

Monadic genetic kernels in Scala
Monadic genetic kernels in ScalaMonadic genetic kernels in Scala
Monadic genetic kernels in Scala
 
Learning Convolutional Neural Networks for Graphs
Learning Convolutional Neural Networks for GraphsLearning Convolutional Neural Networks for Graphs
Learning Convolutional Neural Networks for Graphs
 
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...
Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks, arXiv e-...
 
Anomaly detection using deep one class classifier
Anomaly detection using deep one class classifierAnomaly detection using deep one class classifier
Anomaly detection using deep one class classifier
 
Clustering.pptx
Clustering.pptxClustering.pptx
Clustering.pptx
 
Tuning of PID, SVFB and LQ Controllers Using Genetic Algorithms
Tuning of PID, SVFB and LQ Controllers Using Genetic AlgorithmsTuning of PID, SVFB and LQ Controllers Using Genetic Algorithms
Tuning of PID, SVFB and LQ Controllers Using Genetic Algorithms
 
The world of loss function
The world of loss functionThe world of loss function
The world of loss function
 
SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15
 
Learning LWF Chain Graphs: A Markov Blanket Discovery Approach
Learning LWF Chain Graphs: A Markov Blanket Discovery ApproachLearning LWF Chain Graphs: A Markov Blanket Discovery Approach
Learning LWF Chain Graphs: A Markov Blanket Discovery Approach
 
Efficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketchingEfficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketching
 
TAME: Trainable Attention Mechanism for Explanations
TAME: Trainable Attention Mechanism for ExplanationsTAME: Trainable Attention Mechanism for Explanations
TAME: Trainable Attention Mechanism for Explanations
 
Pr045 deep lab_semantic_segmentation
Pr045 deep lab_semantic_segmentationPr045 deep lab_semantic_segmentation
Pr045 deep lab_semantic_segmentation
 
Barker_SIAMCSE15
Barker_SIAMCSE15Barker_SIAMCSE15
Barker_SIAMCSE15
 
lec26.pptx
lec26.pptxlec26.pptx
lec26.pptx
 
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHMADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
 
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHMADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
ADAPTIVE FUZZY KERNEL CLUSTERING ALGORITHM
 
Inside LoLA - Experiences from building a state space tool for place transiti...
Inside LoLA - Experiences from building a state space tool for place transiti...Inside LoLA - Experiences from building a state space tool for place transiti...
Inside LoLA - Experiences from building a state space tool for place transiti...
 
GKEL_IGARSS_2011.ppt
GKEL_IGARSS_2011.pptGKEL_IGARSS_2011.ppt
GKEL_IGARSS_2011.ppt
 
VJAI Paper Reading#3-KDD2019-ClusterGCN
VJAI Paper Reading#3-KDD2019-ClusterGCNVJAI Paper Reading#3-KDD2019-ClusterGCN
VJAI Paper Reading#3-KDD2019-ClusterGCN
 
Incremental and parallel computation of structural graph summaries for evolvi...
Incremental and parallel computation of structural graph summaries for evolvi...Incremental and parallel computation of structural graph summaries for evolvi...
Incremental and parallel computation of structural graph summaries for evolvi...
 

Graph Kernelpdf

  • 2. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 3. Kernels  A kernel is a function of two objects that quantifies their similarity. Any kernel methods solution comprises two parts: a module that performs the mapping into the embedding or feature space and a learning algorithm designed to discover linear patterns in that space. K(X,X’)=(Φ(X),Φ(X’)) Where Φ is the feature space transformation. Property: Symmetric Positive-Definite C’KC≥0; Graph Kernel
  • 4. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 5. Graph Kernel Graph kernels are used to find similarity between two graphs. It exploits graph topology but, restricts itself to comparing substructure of graph that are computable in polynomial time. Property: Symmetric positive-Definite. Many Graph Kernels have been defined, we will focus on mainly following: 1. Shortest Path Kernel 2. Graphlet Kernel 3. Weisfeiler-Lehman Kernel. GRAPH KERNEL
  • 6. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 7. Shortest Path Kernel Task: Find the shortest path of all lengths among the graphs. Working: 1. Floyd Transformation of graph. 2. Apply shortest path kernel over that transformed graph. Complexity: O(n4) GRAPH KERNEL
  • 8. SP Kernel(Cont.) Floyd Transformation:  Contains same number of nodes as input graph.  Unlike, there exist an edge between nodes in transformed graph if there exist a walk between the nodes in input graph.  Edge are marked as shortest distance between nodes  Complexity O(n3) Floyd-Warshall algorithm GRAPH KERNEL
  • 9. SP Kernel(Cont.) GRAPH KERNEL Kernel: Let S1(E1,V1), S2(E2,V2) be two graphs
  • 11. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 12. Graphlet Kernel Task: Count number of graphlet because isomorphic graphs have same graphlet distribution. Count the number of graphlets of specific size, for linear computation. There are possible graphlet of size k. Let G and G’ be 2 graphs and S and S’ be their respective subgraphs. Complexity: O(ndk-1)  d=maximum degree of node  k= size of graphlet GRAPH KERNEL
  • 13. Graphlet Kernel  computing 3-graphlet kernel for connected component (triangle, V shapes). GRAPH KERNEL
  • 14. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 16. WL Algorithm :iteration 1 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort GRAPH KERNEL
  • 17. WL Algorithm :iteration 1 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort GRAPH KERNEL
  • 18. WL Algorithm :iteration 1 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort GRAPH KERNEL
  • 19. WL Algorithm :iteration 1 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort 3. Relabeling ◦ O(n) GRAPH KERNEL
  • 20. WL Algorithm :iteration 1 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort 3. Relabeling ◦ O(n) 4.Are the labels of G and G’ identical? Yes, continue. GRAPH KERNEL
  • 21. WL Algorithm :iteration 2 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort GRAPH KERNEL
  • 22. WL Algorithm :iteration 2 Each Iteration of WL test comprises of following steps: 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort GRAPH KERNEL
  • 23. WL Algorithm :iteration 2 Each Iteration of WL test comprises of following steps: 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort GRAPH KERNEL
  • 24. WL Algorithm :iteration 2 Each Iteration of WL test comprises of following steps: 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort GRAPH KERNEL
  • 25. WL Algorithm :iteration 2 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort 3. Relabeling ◦ O(n) GRAPH KERNEL
  • 26. WL Algorithm :iteration 2 Each Iteration of WL test comprises of following steps:- 1.Multiset label determination and sorting ◦ O(m) via Bucket Sort 2.label compression ◦ O(m) via Radix Sort 3. Relabeling ◦ O(n) 4.Are the labels of G and G’ identical? NO, output YES. 5.complexity O(hm) for h iteration GRAPH KERNEL
  • 27. WL Kernel Feature{2 3 4 5 6 7 8 9 10} Φ(G1)={3 2 1 0 1 2 2 0 1} Φ(G2)={3 2 1 2 1 0 0 2 1 } K{G1,G2}=16 GRAPH KERNEL
  • 28. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 29. EXPERIMENT 1. INPUT: We have generated 3 different kind of input graphs. a)Edge Removed Graphs: We took a complete graph and removed 10 edges at a time from that graph, iteratively generate new graph. b)Degree Reduced Graph: We took a complete graph and reduced the degree of each node by 1,iteratively generate new graph. c)Node Removed Graph: We took a complete graph and removed 1 node at a time from that graph to generate new graph iteratively. GRAPH KERNEL
  • 33. Shortest Path Kernel norm. GRAPH KERNEL
  • 34. SVM Graph Label Assignment:10 belong to each class, iteratively. 10 Fold SVM Function Used: ‘rbf’ Gamma=0.00001 Accuracy: mean Accuracy output from all 10 folds. GRAPH KERNEL
  • 35. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 39. Outline oKernel oGraph Kernel oShortest Path Kernel oGraphlet Kernel oWL Kernel oExperimentation oResults oConclusion
  • 40. Conclusion Accuracy: Graphlet Kernel > WL Kernel > Shortest Path Kernel I. Graphlet Kernel Exploits the topology very effectively by considering graphlet distribution. II. Shortest Path Kernel shows least accuracy because it computes the number of shortest path while not considering the topology involved in graphs. III. WL Kernel shows moderate performance on our dataset. IV. As we increase number of nodes the accuracy decreases because number of false classification increases. GRAPH KERNEL
  • 41. Conclusion (cont.) Time Graphlet Kernel > Shortest Path Kernel > WL Kernel I. Graphlet Kernel has highest computation time O(ndk-1),since our dataset is pretty dense. II. Shortest Path Kernel O(n4). III. WL Kernel O(Nhm). N=number of Graphs. h=number of iterations. m=number of distinct labels. GRAPH KERNEL