Filters








20,101 Hits in 3.2 sec

Latent Clustering on Graphs with Multiple Edge Types [chapter]

Matthew Rocklin, Ali Pinar
2011 Lecture Notes in Computer Science  
We study clustering on graphs with multiple edge types.  ...  We generalize the concept of clustering in single-edge graphs to multi-edged graphs and discuss how this generates a space of clusterings.  ...  Ongoing work includes more intelligent sampling (intentionally finding distinct clusterings), effects of adding non-linear combinations of edge-types, and searching the space for clusterings with desired  ... 
doi:10.1007/978-3-642-21286-4_4 fatcat:n22twfsjvbewhetqqsbzjiecwa

GraSPy: Graph Statistics in Python [article]

Jaewon Chung, Benjamin D. Pedigo, Eric W. Bridgeford, Bijan K. Varjavand, Hayden S. Helm, Joshua T. Vogelstein
2019 arXiv   pre-print
This package provides flexible and easy-to-use algorithms for analyzing and understanding graphs with a scikit-learn compliant API.  ...  We introduce GraSPy, a Python library devoted to statistical inference, machine learning, and visualization of random graphs and graph populations.  ...  We thank all the contributors for assisting with writing GraSPy. We thank the NeuroData Design class, the NeuroData lab, and Carey E. Priebe at Johns Hopkins University for helpful feedback.  ... 
arXiv:1904.05329v3 fatcat:np5ptkbfn5exvkwplbt7xajprm

On Clustering on Graphs with Multiple Edge Types

Matthew Rocklin, Ali Pinar
2013 Internet Mathematics  
We study clustering on graphs with multiple edge types. Our main motivation is that similarities between objects can be measured by many different metrics.  ...  As such, graphs with multiple edges give a more accurate model to describe similarities between objects than models using single-edge graphs.  ...  We are also grateful to two anonymous reviewers for their helpful comments on an earlier version of this paper.  ... 
doi:10.1080/15427951.2012.678191 fatcat:durcqj6lcjcddnzoi7vzzwwgga

SGVAE: Sequential Graph Variational Autoencoder [article]

Bowen Jing, Ethan A. Chi, Jillian Tang
2019 arXiv   pre-print
In our model, the encoding and decoding of a graph as is framed as a sequential deconstruction and construction process, respectively, enabling the the learning of a latent space.  ...  Generative models of graphs are well-known, but many existing models are limited in scalability and expressivity.  ...  Since we define only one edge type and only one node type, the edge and node initializations in the destructors are constant.  ... 
arXiv:1912.07800v1 fatcat:ff5eig2m6fam5d5fmczwuqj6du

On Clustering on Graphs with Multiple Edge Types [article]

Matthew Rocklin, Ali Pinar
2011 arXiv   pre-print
We study clustering on graphs with multiple edge types. Our main motivation is that similarities between objects can be measured in many different metrics.  ...  As such, graphs with multiple edges is a more accurate model to describe similarities between objects.  ...  Section 4 addresses the latent clustering structure on graphs with multiple edge types, and introduces the concept of meta-clustering.  ... 
arXiv:1109.1605v1 fatcat:hhkh7zl4yjdxfguson3zcxl3w4

Graphs in machine learning: an introduction [article]

Pierre Latouche
2015 arXiv   pre-print
Unsupervised learning algorithms usually aim at visualising graphs in latent spaces and/or clustering the nodes. Both focus on extracting knowledge from graph topologies.  ...  In this context, natural tasks include graph clustering (as in producing clusters of graphs rather than clusters of nodes in a single graph), graph classification, etc. 1 Real networks One of the first  ...  The edges are typed. In parallel, strategies looking for overlapping clusters, where each node can belong to multiple clusters, have been derived.  ... 
arXiv:1506.06962v1 fatcat:qw7ex4vs3fbgbn3c2dppd4kmse

GPSP

Wenyu Du, Shuai Yu, Min Yang, Qiang Qu, Jia Zhu
2018 Companion of the The Web Conference 2018 on The Web Conference 2018 - WWW '18  
In this paper, we propose GPSP, a novel Graph Partition and Space Projection based approach, to learn the representation of a heterogeneous network that consists of multiple types of nodes and links.  ...  Finally, we concatenate the projective vectors from bipartite subnetworks with the ones learned from homogeneous subnetworks to form the final representation of the heterogeneous network.  ...  In bipartite network each edge e 1 2 ∈ E V 1 V 2 connects two different types of nodes 1 ∈ V 1 and 2 ∈ V 2 . Edge-type based graph partition.  ... 
doi:10.1145/3184558.3186928 dblp:conf/www/DuY0QZ18 fatcat:cz3zd6ovzrfvdapuvu5e6e3ksi

Unsupervised generative and graph representation learning for modelling cell differentiation

Ioana Bica, Helena Andrés-Terré, Ana Cvejic, Pietro Liò
2020 Scientific Reports  
We illustrate our methods on datasets from multiple species and also from different sequencing technologies.  ...  Based on these models, we develop a computational framework that consists of methods for identifying the cell types in the dataset, finding driver genes for the differentiation process and obtaining a  ...  (d) Adjacency matrix with predicted links between all cells by Graph-DiffVAE (the colour white indicates a predicted edge).  ... 
doi:10.1038/s41598-020-66166-8 pmid:32555334 fatcat:tqo6mvnk5rhmfgjpzavxorh52y

TRIBAC: Discovering Interpretable Clusters and Latent Structures in Graphs

Jeffrey Chan, Christopher Leckie, James Bailey, Kotagiri Ramamohanarao
2014 2014 IEEE International Conference on Data Mining  
Graph clustering is an important type of approach used to discover these vertex groups and the latent structure within graphs.  ...  One type of approach for graph clustering is nonnegative matrix factorisation However, the formulations of existing factorisation approaches can be overly relaxed and their groupings and results consequently  ...  An important type of analysis to discover these groupings and latent structures is graph clustering, which involves grouping the vertices based on the similarity of their connectivity.  ... 
doi:10.1109/icdm.2014.118 dblp:conf/icdm/ChanLBR14 fatcat:a7lloedrirdg7cjqs5yu3vqszu

Graph Neural Networks for Particle Reconstruction in High Energy Physics detectors [article]

Xiangyang Ju
2020 arXiv   pre-print
Graph Neural Networks (GNNs) are a relatively new class of deep learning architectures which can deal with such data effectively, allowing scientists to incorporate domain knowledge in a graph structure  ...  Reconstruction algorithms identify and measure the kinematic properties of particles produced in high energy collisions and recorded with complex detector systems.  ...  Ongoing work for GNN applications in calorimetry includes studies on how to reconstruct multiple particle types simultaneously using new network architectures which can assign categories to edges.  ... 
arXiv:2003.11603v2 fatcat:5kk4uaae3bebvj5rtxj7mudfri

Visualizing Graph Neural Networks with CorGIE: Corresponding a Graph to Its Embedding [article]

Zipeng Liu, Yang Wang, Jürgen Bernard, Tamara Munzner
2021 arXiv   pre-print
As the key function in CorGIE, we propose the K-hop graph layout to show topological neighbors in hops and their clustering structure.  ...  We propose an approach to corresponding an input graph to its node embedding (aka latent space), a common component of GNNs that is later used for prediction.  ...  ACKNOWLEDGMENTS The authors would like to thank Madison Elliott, Steve Kasica, Michael Oppermann, Ben Shneiderman, and Mara Solen for helpful comments on paper drafts, and the anonymous participants in  ... 
arXiv:2106.12839v2 fatcat:m4nh5jl4gna5bippxx3munxlg4

Network Clustering for Latent State and Changepoint Detection [article]

Madeline Navarro and Genevera I. Allen and Michael Weylandt
2021 arXiv   pre-print
We provide an efficient algorithm for convex network clustering and demonstrate its effectiveness on synthetic examples.  ...  In many situations of interest, however, multiple networks can be constructed to capture different aspects of an underlying phenomenon or to capture changing behavior over time.  ...  Our method may be useful for clustering multiple networks, detecting changepoints in networks over time, or even for detecting latent network temporal states with applications ranging from analyzing social  ... 
arXiv:2111.01273v1 fatcat:vuofgvvlhbbgfjhwkspycbulwm

Agwan: A Generative Model for Labelled, Weighted Graphs [chapter]

Michael Davis, Weiru Liu, Paul Miller, Ruth F. Hunter, Frank Kee
2014 Lecture Notes in Computer Science  
In this paper, we present AGWAN (Attribute Graphs: Weighted and Numeric), a generative model for random graphs with discrete labels and weighted edges.  ...  The model is easily generalised to edges labelled with an arbitrary number of numeric attributes.  ...  Extending AGWAN to multiple attributes We have presented AGWAN for a single discrete vertex label and a single numeric edge label (the weight). Many graphs have multiple labels on vertices and edges.  ... 
doi:10.1007/978-3-319-08407-7_12 fatcat:4uxnbuoz6rf2fe4q3sae7jpxia

Nonparametric Bayesian modeling of complex networks: an introduction

Mikkel N. Schmidt, Morten Morup
2013 IEEE Signal Processing Magazine  
In directed graphs, edges point from one node to another. Edges in a weighed graph have an associated value, e.g., representing the strength of the relation.  ...  directed, weighted, Bipartite, aNd multiple Networks The infinite relational model readily extends to other types of graphs including directed, weighted, and bipartite networks as well as multiple networks  ...  , with many edges joining vertices of the same cluster and comparatively few edges joining vertices of different clusters" [11] .  ... 
doi:10.1109/msp.2012.2235191 fatcat:psujjh4ozrcytbogtue5qtt5ju

Unsupervised generative and graph representation learning for modelling cell differentiation [article]

Ioana Bica, Helena Andres-Terre, Ana Cvejic, Pietro Lio
2019 bioRxiv   pre-print
We illustrate our methods on datasets from multiple species and also from different sequencing technologies.  ...  Based on these models, we develop a computational framework that consists of methods for identifying the cell types in the dataset, finding driver genes for the differentiation process and obtaining a  ...  The encoder in Graph-DiffVAE is represented by a graph convolutional network with multiple layers and with Gaussian output.  ... 
doi:10.1101/801605 fatcat:w73dqzhgafhwpbqmdtfotwu6zm
« Previous Showing results 1 — 15 out of 20,101 results