SlideShare uma empresa Scribd logo
1 de 19
LAB SEMINAR
Nguyen Thanh Sang
Network Science Lab
Dept. of Artificial Intelligence
The Catholic University of Korea
E-mail: sang.ngt99@gmail.com
Improving Graph Neural Network Expressivity via
Subgraph Isomorphism Counting
--- Giorgos Bouritsas , Fabrizio Frasca, Stefanos Zafeiriou, and Michael M.
Bronstein ---
2023-06-01
Content
s
1
⮚ Paper
▪ Introduction
▪ Problem
▪ Contributions
▪ Methodology
▪ Experiments
▪ Conclusion
2
Introduction
 Graph Neural Networks (GNNs) have achieved remarkable results in a variety of applications.
 GNNs use an aggregation function to update the vector representation of each node by transforming and
aggregating the vector representations of its neighbours.
3
Graph Isomorphism
+ Two graphs are also called isomorphic whenever there exists an isomorphism between the two.
+ In graph theory, an isomorphism of graphs 𝐺 and 𝐻
• A bijection between the vertex sets of 𝐺 and H: 𝐹:𝑉(𝐺)→𝑉(𝐻)
• Any two vertices 𝑢 and 𝑣 of 𝐺 are adjacent in 𝐺 if and only if 𝑓(𝑢) and 𝑓(𝑣) are adjacent in 𝐻
4
Graph Automorphism
+ A bijection mapping onto itself
• When 𝐺 and 𝐻 are one and the same graph
• A form of symmetry
+ Problem
• Testing whether a graph has a nontrivial automorphism
=> Computational complexity
• Constructing the automorphism group
=> Orbit
5
Problems
❖ The Weisfeiler-Lehman test: representative test for isomorphism
• Low computational complexity
• Good for all graphs
 Limit in some case
• Not apply in real world data.
 Arbitrarily initialized for test.
Initial 1st iteration 2nd iteration 3rd iteration
6
Problems
• Since message-passing GNNs are at most as powerful as the Weisfeiler Leman test (WL), they
are limited in their abilities to adequately exploit the graph structure, e.g. by counting
substructures.
 important in the study of complex networks.
 How to go beyond isotropic, i.e., locally symmetric, aggregation functions?
 How to ensure e structural characteristics of the graph?
 How to achieve the above two without sacrificing invariance to isomorphism?
7
Contributions
• Break local symmetries by introducing structural information in the aggregation function.
• Each neighbour (message) is transformed differently depending on its structural relationship with
the central node.
 counting the appearance of certain substructures.
• Graph Substructure Network (GSN) is strictly more expressive than traditional GNNs for the vast
majority of substructures, while retaining the locality of message passing, as opposed to higher-
order methods.
• When choosing the structural inductive biases based on domainspecific knowledge, GSN
achieves state-of-the-art results
8
Structural Features
+ Features encoded from structural roles by counting the appearance
of certain substructures.
+ Step 1: A set of small (connected) graphs 𝐻 ∈ ℋ, ℋ = 𝐻1, 𝐻2, ⋯ , 𝐻𝐾,
e.g., cycles, paths, cliques, or trees
- Find its(each graph 𝐻 ∈ ℋ) isomorphic subgraphs in 𝐺 denoted 𝐺𝑆
- For each node 𝑣 ∈ 𝑉𝐺𝑆
, infer its role w.r.t 𝐻 by obtaining the orbit of its mapping 𝑓 𝑣 in 𝐻, Orb𝐻 𝑓 𝑣
+ Step 2: the 𝑣𝑒𝑟𝑡𝑒𝑥 𝑠𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑎𝑙 𝑓𝑒𝑎𝑡𝑢𝑟𝑒 𝐱𝐻
𝑉
𝑣 of 𝑣 by counting all the possible appearances of different orbits in 𝑣:
- For all 𝑖 ∈ 1, 2, ⋯ , 𝑑𝐻 : 𝐱𝐻
𝑉
𝑣 = 𝐺𝑆 ≃ 𝐻 𝑣 ∈ 𝑉𝐺𝑆
, 𝑓 𝑣 ∈ 𝑂𝐻,𝑖
𝑉
the number of elements in the set of nodes used in the orbit that make up a specific isomorphic mapping
𝑓: functions can map a subgraph 𝐺𝑆 to 𝐻
can be used to determine the orbit mapping of each node 𝑣
- Feature vector: 𝐱𝑣
𝑉
= 𝐱𝐻,1
𝑉
𝑣 , 𝐱𝐻,2
𝑉
𝑣 , ⋯ , 𝐱𝐻,𝐾
𝑉
𝑣
- The 𝑒𝑑𝑔𝑒 𝑠𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑎𝑙 𝑓𝑒𝑎𝑡𝑢𝑟𝑒 𝐱𝐻
𝐸
𝑢, 𝑣 of 𝑢, 𝑣 : 𝐱𝐻
𝐸
𝑢, 𝑣 = 𝐺𝑆 ≃ 𝐻 𝑢, 𝑣 ∈ ℰ𝐺𝑆
, 𝑓 𝑢 , 𝑓 𝑣 ∈ 𝑂𝐻,𝑖
𝐸
𝐱𝑢,𝑣
𝐸
= 𝐱𝐻,1
𝐸
𝑢, 𝑣 , 𝐱𝐻,2
𝐸
𝑢, 𝑣 , ⋯ , 𝐱𝐻,𝐾
𝐸
𝑢, 𝑣
9
Structure-aware Message Passing
The substructure layer as a Message Passing Neural Network:
[Message Info.] + [Structural Roles Info.]
𝐡𝑡+1 = UP𝑡+1 𝐡𝑣
𝑡 , 𝐦𝑣
𝑡+1
UP𝑡+1 : an arbitrary function approximator (e.g., an MLP)
𝑀𝑡+1
: the neighborhood aggregation function
An arbitrary function on multisets
𝐞𝑢,𝑣: the edge features
the vertex structural identifiers
the edge structural identifiers
10
Power of GSNs
+ GSN > MPNN: MPNN-based architecture
+ GSN > 1-WL: Considering possible all orbits
+ Open Problem: the fixed subgraph has not been
defined yet
Rook's 4x4 graph Shrikhande graph
(4-clique) (triangle)
2-FWL fails
11
Experiments
Settings
+ Baseline: MPNN with MLP
+ Substructure families: Cycles, paths, trees and cliques
+ Substructure size: k
+ Datasets: Synthetic, TUD, ZINC, and OGB-MOLHIV
12
Synthetic Graph Isomorphism Test
+ Dataset: a collection of Strongly Regular graphs of size up to 35 nodes
Isomorphic decision
+ The Euclidean distance of their representations is smaller than a predefined threshold 𝜖.
+ The number of failure cases of GSN decreases rapidly as we increase k; cycles and
paths of maximum length k = 6.
13
TUD Graph Classification
• Dataset: Bioinformatics, Social networks
• Comparison: GNN, Graph Kernels with 10-fold cv
• Base architecture: GIN
• Best performing substructures both for GSN-e and GSN-v
=> The proposed model obtains SOTA performance in most of the datasets, with a considerable margin
against the main GNN baselines in some cases.
14
ZINC Molecular Graphs
• Dataset
+ Commercially-available compounds for
virtual screening
+ John J. Irwin et al.
+ Graph regression (mainly)
• Task
+ k-cycle counting
+ Molecule: 10k / 2k
+ Regression (MAE)
=> GSN achieves state-of-the-art results
outperforming all the baseline architectures.
15
OGB-MOLHIV
• GSN seamlessly improves the performance of the base architecture
• Cyclical substructures are a good inductive bias when learning on molecules, confirming our
results on the ZINC dataset, while the same holds for triangles in PPA networks. Tasks defined on
graphs with community structure correlate with the presence of triangles (or cliques), as was the
case for social networks in the TU Datasets experiments.
• General purpose GNNs benefit from symmetry breaking mechanisms, either in the form of
eigenvectors (DGN) or in the form of substructures.
16
Ablation Studies
• The test error is not guaranteed to decrease when the identifiers become more discriminative.
• This method fails to improve the baseline architecture in terms of the performance in the test set.
 unique identifiers can be hard to generalise when chosen in a non-permutation equivariant way and
motivates once more the importance of choosing the identifiers not only based on their
discriminative power, but also in a way that allows incorporating the appropriate inductive biases.
• GSN manages to generalize much better even with a small fraction of the training dataset.
17
Conclusions
• A novel way to design structure aware graph neural networks. Motivated by the limitations of traditional
GNNs to capture important topological properties of the graph.
• A message passing scheme enhanced with structural features that are extracted by subgraph
isomorphism.
• For some types of substructures such as paths and cycles the counting can be done with significantly
lower complexity.
• The computationally expensive step is done only once as preprocessing and thus does not affect
network training and inference that remain linear, the same way as in message-passing neural
networks. The memory complexity in training and inference is linear as well
• Most importantly, the expressive power of GSN is different from k-WL tests and in some cases is
stronger
18
Thank you!

Mais conteúdo relacionado

Semelhante a NS-CUK Seminar: S.T.Nguyen, Review on "Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting", IEEE 2020

Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...
Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...
Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...
Ashray Bhandare
 

Semelhante a NS-CUK Seminar: S.T.Nguyen, Review on "Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting", IEEE 2020 (20)

Characteristics of Networks Generated by Kernel Growing Neural Gas
Characteristics of Networks Generated by Kernel Growing Neural GasCharacteristics of Networks Generated by Kernel Growing Neural Gas
Characteristics of Networks Generated by Kernel Growing Neural Gas
 
CHARACTERISTICS OF NETWORKS GENERATED BY KERNEL GROWING NEURAL GAS
CHARACTERISTICS OF NETWORKS GENERATED BY KERNEL GROWING NEURAL GASCHARACTERISTICS OF NETWORKS GENERATED BY KERNEL GROWING NEURAL GAS
CHARACTERISTICS OF NETWORKS GENERATED BY KERNEL GROWING NEURAL GAS
 
NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...
NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...
NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...
 
NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...
NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...
NS-CUK Seminar: J.H.Lee, Review on "Learnable Structural Semantic Readout for...
 
NS-CUK Seminar: S.T.Nguyen Review on "Accurate learning of graph representati...
NS-CUK Seminar: S.T.Nguyen Review on "Accurate learning of graph representati...NS-CUK Seminar: S.T.Nguyen Review on "Accurate learning of graph representati...
NS-CUK Seminar: S.T.Nguyen Review on "Accurate learning of graph representati...
 
NS-CUK Seminar: S.T.Nguyen, Review on "Geom-GCN: Geometric Graph Convolutiona...
NS-CUK Seminar: S.T.Nguyen, Review on "Geom-GCN: Geometric Graph Convolutiona...NS-CUK Seminar: S.T.Nguyen, Review on "Geom-GCN: Geometric Graph Convolutiona...
NS-CUK Seminar: S.T.Nguyen, Review on "Geom-GCN: Geometric Graph Convolutiona...
 
NS - CUK Seminar: S.T.Nguyen, Review on "Hypergraph Neural Networks", AAAI 2019
NS - CUK Seminar: S.T.Nguyen, Review on "Hypergraph Neural Networks", AAAI 2019NS - CUK Seminar: S.T.Nguyen, Review on "Hypergraph Neural Networks", AAAI 2019
NS - CUK Seminar: S.T.Nguyen, Review on "Hypergraph Neural Networks", AAAI 2019
 
240325_JW_labseminar[node2vec: Scalable Feature Learning for Networks].pptx
240325_JW_labseminar[node2vec: Scalable Feature Learning for Networks].pptx240325_JW_labseminar[node2vec: Scalable Feature Learning for Networks].pptx
240325_JW_labseminar[node2vec: Scalable Feature Learning for Networks].pptx
 
Exploring Randomly Wired Neural Networks for Image Recognition
Exploring Randomly Wired Neural Networks for Image RecognitionExploring Randomly Wired Neural Networks for Image Recognition
Exploring Randomly Wired Neural Networks for Image Recognition
 
NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Transformer with Ad...
NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Transformer with Ad...NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Transformer with Ad...
NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Transformer with Ad...
 
240401_Thuy_Labseminar[Train Once and Explain Everywhere: Pre-training Interp...
240401_Thuy_Labseminar[Train Once and Explain Everywhere: Pre-training Interp...240401_Thuy_Labseminar[Train Once and Explain Everywhere: Pre-training Interp...
240401_Thuy_Labseminar[Train Once and Explain Everywhere: Pre-training Interp...
 
Spectral cnn
Spectral cnnSpectral cnn
Spectral cnn
 
NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Convolutional Netwo...
NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Convolutional Netwo...NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Convolutional Netwo...
NS-CUK Seminar: S.T.Nguyen, Review on "Hierarchical Graph Convolutional Netwo...
 
Colloquium.pptx
Colloquium.pptxColloquium.pptx
Colloquium.pptx
 
Applications of machine learning in Wireless sensor networks.
Applications of machine learning in Wireless sensor networks.Applications of machine learning in Wireless sensor networks.
Applications of machine learning in Wireless sensor networks.
 
Learning Convolutional Neural Networks for Graphs
Learning Convolutional Neural Networks for GraphsLearning Convolutional Neural Networks for Graphs
Learning Convolutional Neural Networks for Graphs
 
[20240422_LabSeminar_Huy]Taming_Effect.pptx
[20240422_LabSeminar_Huy]Taming_Effect.pptx[20240422_LabSeminar_Huy]Taming_Effect.pptx
[20240422_LabSeminar_Huy]Taming_Effect.pptx
 
230727_HB_JointJournalClub.pptx
230727_HB_JointJournalClub.pptx230727_HB_JointJournalClub.pptx
230727_HB_JointJournalClub.pptx
 
Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...
Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...
Bio-inspired Algorithms for Evolving the Architecture of Convolutional Neural...
 
Multidimensional RNN
Multidimensional RNNMultidimensional RNN
Multidimensional RNN
 

Mais de ssuser4b1f48

Mais de ssuser4b1f48 (20)

NS-CUK Seminar: V.T.Hoang, Review on "GOAT: A Global Transformer on Large-sca...
NS-CUK Seminar: V.T.Hoang, Review on "GOAT: A Global Transformer on Large-sca...NS-CUK Seminar: V.T.Hoang, Review on "GOAT: A Global Transformer on Large-sca...
NS-CUK Seminar: V.T.Hoang, Review on "GOAT: A Global Transformer on Large-sca...
 
NS-CUK Seminar: J.H.Lee, Review on "Graph Propagation Transformer for Graph R...
NS-CUK Seminar: J.H.Lee, Review on "Graph Propagation Transformer for Graph R...NS-CUK Seminar: J.H.Lee, Review on "Graph Propagation Transformer for Graph R...
NS-CUK Seminar: J.H.Lee, Review on "Graph Propagation Transformer for Graph R...
 
NS-CUK Seminar: H.B.Kim, Review on "Cluster-GCN: An Efficient Algorithm for ...
NS-CUK Seminar: H.B.Kim,  Review on "Cluster-GCN: An Efficient Algorithm for ...NS-CUK Seminar: H.B.Kim,  Review on "Cluster-GCN: An Efficient Algorithm for ...
NS-CUK Seminar: H.B.Kim, Review on "Cluster-GCN: An Efficient Algorithm for ...
 
NS-CUK Seminar: H.E.Lee, Review on "Weisfeiler and Leman Go Neural: Higher-O...
NS-CUK Seminar: H.E.Lee,  Review on "Weisfeiler and Leman Go Neural: Higher-O...NS-CUK Seminar: H.E.Lee,  Review on "Weisfeiler and Leman Go Neural: Higher-O...
NS-CUK Seminar: H.E.Lee, Review on "Weisfeiler and Leman Go Neural: Higher-O...
 
NS-CUK Seminar:V.T.Hoang, Review on "GRPE: Relative Positional Encoding for G...
NS-CUK Seminar:V.T.Hoang, Review on "GRPE: Relative Positional Encoding for G...NS-CUK Seminar:V.T.Hoang, Review on "GRPE: Relative Positional Encoding for G...
NS-CUK Seminar:V.T.Hoang, Review on "GRPE: Relative Positional Encoding for G...
 
Aug 22nd, 2023: Case Studies - The Art and Science of Animation Production)
Aug 22nd, 2023: Case Studies - The Art and Science of Animation Production)Aug 22nd, 2023: Case Studies - The Art and Science of Animation Production)
Aug 22nd, 2023: Case Studies - The Art and Science of Animation Production)
 
Aug 17th, 2023: Case Studies - Examining Gamification through Virtual/Augment...
Aug 17th, 2023: Case Studies - Examining Gamification through Virtual/Augment...Aug 17th, 2023: Case Studies - Examining Gamification through Virtual/Augment...
Aug 17th, 2023: Case Studies - Examining Gamification through Virtual/Augment...
 
Aug 10th, 2023: Case Studies - The Power of eXtended Reality (XR) with 360°
Aug 10th, 2023: Case Studies - The Power of eXtended Reality (XR) with 360°Aug 10th, 2023: Case Studies - The Power of eXtended Reality (XR) with 360°
Aug 10th, 2023: Case Studies - The Power of eXtended Reality (XR) with 360°
 
Aug 8th, 2023: Case Studies - Utilizing eXtended Reality (XR) in Drones)
Aug 8th, 2023: Case Studies - Utilizing eXtended Reality (XR) in Drones)Aug 8th, 2023: Case Studies - Utilizing eXtended Reality (XR) in Drones)
Aug 8th, 2023: Case Studies - Utilizing eXtended Reality (XR) in Drones)
 
NS-CUK Seminar: H.E.Lee, Review on "Gated Graph Sequence Neural Networks", I...
NS-CUK Seminar: H.E.Lee,  Review on "Gated Graph Sequence Neural Networks", I...NS-CUK Seminar: H.E.Lee,  Review on "Gated Graph Sequence Neural Networks", I...
NS-CUK Seminar: H.E.Lee, Review on "Gated Graph Sequence Neural Networks", I...
 
NS-CUK Seminar:V.T.Hoang, Review on "Augmentation-Free Self-Supervised Learni...
NS-CUK Seminar:V.T.Hoang, Review on "Augmentation-Free Self-Supervised Learni...NS-CUK Seminar:V.T.Hoang, Review on "Augmentation-Free Self-Supervised Learni...
NS-CUK Seminar:V.T.Hoang, Review on "Augmentation-Free Self-Supervised Learni...
 
NS-CUK Journal club: H.E.Lee, Review on " A biomedical knowledge graph-based ...
NS-CUK Journal club: H.E.Lee, Review on " A biomedical knowledge graph-based ...NS-CUK Journal club: H.E.Lee, Review on " A biomedical knowledge graph-based ...
NS-CUK Journal club: H.E.Lee, Review on " A biomedical knowledge graph-based ...
 
NS-CUK Seminar: H.E.Lee, Review on "PTE: Predictive Text Embedding through L...
NS-CUK Seminar: H.E.Lee,  Review on "PTE: Predictive Text Embedding through L...NS-CUK Seminar: H.E.Lee,  Review on "PTE: Predictive Text Embedding through L...
NS-CUK Seminar: H.E.Lee, Review on "PTE: Predictive Text Embedding through L...
 
NS-CUK Seminar: H.B.Kim, Review on "Inductive Representation Learning on Lar...
NS-CUK Seminar: H.B.Kim,  Review on "Inductive Representation Learning on Lar...NS-CUK Seminar: H.B.Kim,  Review on "Inductive Representation Learning on Lar...
NS-CUK Seminar: H.B.Kim, Review on "Inductive Representation Learning on Lar...
 
NS-CUK Seminar: H.E.Lee, Review on "PTE: Predictive Text Embedding through L...
NS-CUK Seminar: H.E.Lee,  Review on "PTE: Predictive Text Embedding through L...NS-CUK Seminar: H.E.Lee,  Review on "PTE: Predictive Text Embedding through L...
NS-CUK Seminar: H.E.Lee, Review on "PTE: Predictive Text Embedding through L...
 
NS-CUK Seminar: J.H.Lee, Review on "Relational Self-Supervised Learning on Gr...
NS-CUK Seminar: J.H.Lee, Review on "Relational Self-Supervised Learning on Gr...NS-CUK Seminar: J.H.Lee, Review on "Relational Self-Supervised Learning on Gr...
NS-CUK Seminar: J.H.Lee, Review on "Relational Self-Supervised Learning on Gr...
 
NS-CUK Seminar: H.B.Kim, Review on "metapath2vec: Scalable representation le...
NS-CUK Seminar: H.B.Kim,  Review on "metapath2vec: Scalable representation le...NS-CUK Seminar: H.B.Kim,  Review on "metapath2vec: Scalable representation le...
NS-CUK Seminar: H.B.Kim, Review on "metapath2vec: Scalable representation le...
 
NS-CUK Seminar: H.E.Lee, Review on "Graph Star Net for Generalized Multi-Tas...
NS-CUK Seminar: H.E.Lee,  Review on "Graph Star Net for Generalized Multi-Tas...NS-CUK Seminar: H.E.Lee,  Review on "Graph Star Net for Generalized Multi-Tas...
NS-CUK Seminar: H.E.Lee, Review on "Graph Star Net for Generalized Multi-Tas...
 
NS-CUK Seminar: V.T.Hoang, Review on "Namkyeong Lee, et al. Relational Self-...
NS-CUK Seminar:  V.T.Hoang, Review on "Namkyeong Lee, et al. Relational Self-...NS-CUK Seminar:  V.T.Hoang, Review on "Namkyeong Lee, et al. Relational Self-...
NS-CUK Seminar: V.T.Hoang, Review on "Namkyeong Lee, et al. Relational Self-...
 
NS-CUK Seminar: H.B.Kim, Review on "metapath2vec: Scalable representation l...
NS-CUK Seminar:  H.B.Kim,  Review on "metapath2vec: Scalable representation l...NS-CUK Seminar:  H.B.Kim,  Review on "metapath2vec: Scalable representation l...
NS-CUK Seminar: H.B.Kim, Review on "metapath2vec: Scalable representation l...
 

Último

Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Victor Rentea
 
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
?#DUbAI#??##{{(☎️+971_581248768%)**%*]'#abortion pills for sale in dubai@
 
Architecting Cloud Native Applications
Architecting Cloud Native ApplicationsArchitecting Cloud Native Applications
Architecting Cloud Native Applications
WSO2
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Safe Software
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire business
panagenda
 

Último (20)

ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
Emergent Methods: Multi-lingual narrative tracking in the news - real-time ex...
Emergent Methods: Multi-lingual narrative tracking in the news - real-time ex...Emergent Methods: Multi-lingual narrative tracking in the news - real-time ex...
Emergent Methods: Multi-lingual narrative tracking in the news - real-time ex...
 
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin WoodPolkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
 
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
 
Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024Finding Java's Hidden Performance Traps @ DevoxxUK 2024
Finding Java's Hidden Performance Traps @ DevoxxUK 2024
 
CNIC Information System with Pakdata Cf In Pakistan
CNIC Information System with Pakdata Cf In PakistanCNIC Information System with Pakdata Cf In Pakistan
CNIC Information System with Pakdata Cf In Pakistan
 
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
 
Manulife - Insurer Transformation Award 2024
Manulife - Insurer Transformation Award 2024Manulife - Insurer Transformation Award 2024
Manulife - Insurer Transformation Award 2024
 
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdfRising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
 
EMPOWERMENT TECHNOLOGY GRADE 11 QUARTER 2 REVIEWER
EMPOWERMENT TECHNOLOGY GRADE 11 QUARTER 2 REVIEWEREMPOWERMENT TECHNOLOGY GRADE 11 QUARTER 2 REVIEWER
EMPOWERMENT TECHNOLOGY GRADE 11 QUARTER 2 REVIEWER
 
Architecting Cloud Native Applications
Architecting Cloud Native ApplicationsArchitecting Cloud Native Applications
Architecting Cloud Native Applications
 
Cyberprint. Dark Pink Apt Group [EN].pdf
Cyberprint. Dark Pink Apt Group [EN].pdfCyberprint. Dark Pink Apt Group [EN].pdf
Cyberprint. Dark Pink Apt Group [EN].pdf
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
 
FWD Group - Insurer Innovation Award 2024
FWD Group - Insurer Innovation Award 2024FWD Group - Insurer Innovation Award 2024
FWD Group - Insurer Innovation Award 2024
 
Exploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with MilvusExploring Multimodal Embeddings with Milvus
Exploring Multimodal Embeddings with Milvus
 
Corporate and higher education May webinar.pptx
Corporate and higher education May webinar.pptxCorporate and higher education May webinar.pptx
Corporate and higher education May webinar.pptx
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire business
 

NS-CUK Seminar: S.T.Nguyen, Review on "Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting", IEEE 2020

  • 1. LAB SEMINAR Nguyen Thanh Sang Network Science Lab Dept. of Artificial Intelligence The Catholic University of Korea E-mail: sang.ngt99@gmail.com Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting --- Giorgos Bouritsas , Fabrizio Frasca, Stefanos Zafeiriou, and Michael M. Bronstein --- 2023-06-01
  • 2. Content s 1 ⮚ Paper ▪ Introduction ▪ Problem ▪ Contributions ▪ Methodology ▪ Experiments ▪ Conclusion
  • 3. 2 Introduction  Graph Neural Networks (GNNs) have achieved remarkable results in a variety of applications.  GNNs use an aggregation function to update the vector representation of each node by transforming and aggregating the vector representations of its neighbours.
  • 4. 3 Graph Isomorphism + Two graphs are also called isomorphic whenever there exists an isomorphism between the two. + In graph theory, an isomorphism of graphs 𝐺 and 𝐻 • A bijection between the vertex sets of 𝐺 and H: 𝐹:𝑉(𝐺)→𝑉(𝐻) • Any two vertices 𝑢 and 𝑣 of 𝐺 are adjacent in 𝐺 if and only if 𝑓(𝑢) and 𝑓(𝑣) are adjacent in 𝐻
  • 5. 4 Graph Automorphism + A bijection mapping onto itself • When 𝐺 and 𝐻 are one and the same graph • A form of symmetry + Problem • Testing whether a graph has a nontrivial automorphism => Computational complexity • Constructing the automorphism group => Orbit
  • 6. 5 Problems ❖ The Weisfeiler-Lehman test: representative test for isomorphism • Low computational complexity • Good for all graphs  Limit in some case • Not apply in real world data.  Arbitrarily initialized for test. Initial 1st iteration 2nd iteration 3rd iteration
  • 7. 6 Problems • Since message-passing GNNs are at most as powerful as the Weisfeiler Leman test (WL), they are limited in their abilities to adequately exploit the graph structure, e.g. by counting substructures.  important in the study of complex networks.  How to go beyond isotropic, i.e., locally symmetric, aggregation functions?  How to ensure e structural characteristics of the graph?  How to achieve the above two without sacrificing invariance to isomorphism?
  • 8. 7 Contributions • Break local symmetries by introducing structural information in the aggregation function. • Each neighbour (message) is transformed differently depending on its structural relationship with the central node.  counting the appearance of certain substructures. • Graph Substructure Network (GSN) is strictly more expressive than traditional GNNs for the vast majority of substructures, while retaining the locality of message passing, as opposed to higher- order methods. • When choosing the structural inductive biases based on domainspecific knowledge, GSN achieves state-of-the-art results
  • 9. 8 Structural Features + Features encoded from structural roles by counting the appearance of certain substructures. + Step 1: A set of small (connected) graphs 𝐻 ∈ ℋ, ℋ = 𝐻1, 𝐻2, ⋯ , 𝐻𝐾, e.g., cycles, paths, cliques, or trees - Find its(each graph 𝐻 ∈ ℋ) isomorphic subgraphs in 𝐺 denoted 𝐺𝑆 - For each node 𝑣 ∈ 𝑉𝐺𝑆 , infer its role w.r.t 𝐻 by obtaining the orbit of its mapping 𝑓 𝑣 in 𝐻, Orb𝐻 𝑓 𝑣 + Step 2: the 𝑣𝑒𝑟𝑡𝑒𝑥 𝑠𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑎𝑙 𝑓𝑒𝑎𝑡𝑢𝑟𝑒 𝐱𝐻 𝑉 𝑣 of 𝑣 by counting all the possible appearances of different orbits in 𝑣: - For all 𝑖 ∈ 1, 2, ⋯ , 𝑑𝐻 : 𝐱𝐻 𝑉 𝑣 = 𝐺𝑆 ≃ 𝐻 𝑣 ∈ 𝑉𝐺𝑆 , 𝑓 𝑣 ∈ 𝑂𝐻,𝑖 𝑉 the number of elements in the set of nodes used in the orbit that make up a specific isomorphic mapping 𝑓: functions can map a subgraph 𝐺𝑆 to 𝐻 can be used to determine the orbit mapping of each node 𝑣 - Feature vector: 𝐱𝑣 𝑉 = 𝐱𝐻,1 𝑉 𝑣 , 𝐱𝐻,2 𝑉 𝑣 , ⋯ , 𝐱𝐻,𝐾 𝑉 𝑣 - The 𝑒𝑑𝑔𝑒 𝑠𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑎𝑙 𝑓𝑒𝑎𝑡𝑢𝑟𝑒 𝐱𝐻 𝐸 𝑢, 𝑣 of 𝑢, 𝑣 : 𝐱𝐻 𝐸 𝑢, 𝑣 = 𝐺𝑆 ≃ 𝐻 𝑢, 𝑣 ∈ ℰ𝐺𝑆 , 𝑓 𝑢 , 𝑓 𝑣 ∈ 𝑂𝐻,𝑖 𝐸 𝐱𝑢,𝑣 𝐸 = 𝐱𝐻,1 𝐸 𝑢, 𝑣 , 𝐱𝐻,2 𝐸 𝑢, 𝑣 , ⋯ , 𝐱𝐻,𝐾 𝐸 𝑢, 𝑣
  • 10. 9 Structure-aware Message Passing The substructure layer as a Message Passing Neural Network: [Message Info.] + [Structural Roles Info.] 𝐡𝑡+1 = UP𝑡+1 𝐡𝑣 𝑡 , 𝐦𝑣 𝑡+1 UP𝑡+1 : an arbitrary function approximator (e.g., an MLP) 𝑀𝑡+1 : the neighborhood aggregation function An arbitrary function on multisets 𝐞𝑢,𝑣: the edge features the vertex structural identifiers the edge structural identifiers
  • 11. 10 Power of GSNs + GSN > MPNN: MPNN-based architecture + GSN > 1-WL: Considering possible all orbits + Open Problem: the fixed subgraph has not been defined yet Rook's 4x4 graph Shrikhande graph (4-clique) (triangle) 2-FWL fails
  • 12. 11 Experiments Settings + Baseline: MPNN with MLP + Substructure families: Cycles, paths, trees and cliques + Substructure size: k + Datasets: Synthetic, TUD, ZINC, and OGB-MOLHIV
  • 13. 12 Synthetic Graph Isomorphism Test + Dataset: a collection of Strongly Regular graphs of size up to 35 nodes Isomorphic decision + The Euclidean distance of their representations is smaller than a predefined threshold 𝜖. + The number of failure cases of GSN decreases rapidly as we increase k; cycles and paths of maximum length k = 6.
  • 14. 13 TUD Graph Classification • Dataset: Bioinformatics, Social networks • Comparison: GNN, Graph Kernels with 10-fold cv • Base architecture: GIN • Best performing substructures both for GSN-e and GSN-v => The proposed model obtains SOTA performance in most of the datasets, with a considerable margin against the main GNN baselines in some cases.
  • 15. 14 ZINC Molecular Graphs • Dataset + Commercially-available compounds for virtual screening + John J. Irwin et al. + Graph regression (mainly) • Task + k-cycle counting + Molecule: 10k / 2k + Regression (MAE) => GSN achieves state-of-the-art results outperforming all the baseline architectures.
  • 16. 15 OGB-MOLHIV • GSN seamlessly improves the performance of the base architecture • Cyclical substructures are a good inductive bias when learning on molecules, confirming our results on the ZINC dataset, while the same holds for triangles in PPA networks. Tasks defined on graphs with community structure correlate with the presence of triangles (or cliques), as was the case for social networks in the TU Datasets experiments. • General purpose GNNs benefit from symmetry breaking mechanisms, either in the form of eigenvectors (DGN) or in the form of substructures.
  • 17. 16 Ablation Studies • The test error is not guaranteed to decrease when the identifiers become more discriminative. • This method fails to improve the baseline architecture in terms of the performance in the test set.  unique identifiers can be hard to generalise when chosen in a non-permutation equivariant way and motivates once more the importance of choosing the identifiers not only based on their discriminative power, but also in a way that allows incorporating the appropriate inductive biases. • GSN manages to generalize much better even with a small fraction of the training dataset.
  • 18. 17 Conclusions • A novel way to design structure aware graph neural networks. Motivated by the limitations of traditional GNNs to capture important topological properties of the graph. • A message passing scheme enhanced with structural features that are extracted by subgraph isomorphism. • For some types of substructures such as paths and cycles the counting can be done with significantly lower complexity. • The computationally expensive step is done only once as preprocessing and thus does not affect network training and inference that remain linear, the same way as in message-passing neural networks. The memory complexity in training and inference is linear as well • Most importantly, the expressive power of GSN is different from k-WL tests and in some cases is stronger