This is a list of resources, links and papers on the use of intersection of Graph Theory, Analysis and Data within Machine Learning, and vice versa.
Graphs are useful in machine learning problems, machine learning is useful in the analysis of Graphs and data sets that can be represented as graphs.
Graph data is becoming all the more prevalent and indeed with the advent of graph data bases, we can represent and query much of our relational sata as a graph giving us a new branch of analytics rto wield at problems as diverse as customer segmentation to tracking disease mutation though a population.
In the following sections there can be quite a lot of overlap.We've use judgement to categorise some of these papers and references. The field is evolving rapidly and we're constantly finiding new references that are difficult to categorise.
This list is curated and includes links to what we feel are some of teh key influencing parts of research. We'll be growing the list and there is plently more out there should to you go looking. In the meantime, these are good points of reference.
- Graph Classification
- Classification of data/entities based on the nature of their Graph, each full (sub-)graph is assigned a class label
- Data / Application Spaces
- Predictive toxocology
- References
- Vertex Classification
- Within a graph, each vertex is assigned a class label
- Graph Based Similarity
- Determining Graph Isomorphisms are hard (NP-complete/hard? what's teh complexity here), we often need to approximate, we can take ML based approaches to do so.
- use vector based classification via the kernel trick, SVMs?
- Frequent Subgraph Mining
- Explanation - http://data-mining.philippe-fournier-viger.com/introduction-frequent-subgraph-mining/
- Boosting with gBoost - https://link.springer.com/article/10.1007/s10994-008-5089-z
- Graph based Anomaly Detection
- a survey paper: https://link.springer.com/article/10.1007%2Fs10618-014-0365-y
-
Community Detection with Graph Neural Networks
-
Link Prediction in Convolutional Neural Networks
-
Modularity based Community Detection with Deep Learning
-
Inductive Respresentation Learning on Large Graphs RIGHT CATEGORY??
-
Spectral Features
-
Graph Based Semi and Unsupervised Classification and Segmentation of Microscopic Images (regularization)
-
node2vec - Scalable Feature Learning for Networks [maybe demoable]
description: node2vec is an algorithmic framework for representational learning on graphs. Given any graph, it can learn continuous feature representations for the nodes, which can then be used for various downstream machine learning tasks.
-
graph2vec - Learning Distributed Representations of Graphs [maybe demoable]
-
DeepWalk - Online Learning of Social Representations DeepWalk, a novel approach for learning latent representations of vertices in a network. These latent representations encode social relations in a continuous vector space, which is easily exploited by statistical models. DeepWalk generalizes recent advancements in language modeling and unsupervised feature learning (or deep learning) from sequences of words to graphs. DeepWalk uses local information obtained from truncated random walks to learn latent representations by treating walks as the equivalent of sentences. We demonstrate DeepWalk's latent representations on several multi-label network classification tasks for social networks such as BlogCatalog, Flickr, and YouTube. Our results show that DeepWalk outperforms challenging baselines which are allowed a global view of the network, especially in the presence of missing information. Unweighted graphs.
-
Learning Graph Representations Learning vertex representations of Graph Structure
-
Learning deep representations for graph clustering Use deep neural networks (stacked autoencoder) to learn features
-
(Deep Neural Networks for Learning Graph Representations) Learn low dimensional representation on a vertex that encodes the local structure of the graph. Could also be described as graph embeddings.
-
-
Variational Graph Auto-Encoders
-
DeepGraph
-
Graph Kernels an Introduction
-
Transfer Learning for Deep Learning on Graph-Structured Data
- paper
-
CayleyNets: Graph Convolutional Neural Networks with Complex Rational Spectral Filters
-
Geometric Deep Learning
-
Classifying Graphs as Images with Convolutional Neural Networks - paper
-
Graph Convolutional Networks - Thomas Kipf
-
Book - Introduction to Semi-Supervised Learning
-
(book chapter on based-based)[https://books.google.es/books?id=c_haJrQ0ScAC&pg=PA44&lpg=PA44&dq=learn+graph+edge+weighting&source=bl&ots=KY2SR00zjX&sig=jA0Vfp7bgEVjsOpjZJ7gqx6CYDc&hl=en&sa=X&ved=0ahUKEwiHmLnKg4HWAhUGZlAKHZ81ASo4ChDoAQhEMAU#v=onepage&q=learn%20graph%20edge%20weighting&f=false]
-
- Graph based semi-supervised learning
- Active graph-based semi supervised learning
-
Revisiting Semi-Supervised Learning with Graph Embeddings
-
Semi-Supervised Learning with graphs
-
Spectrum-based deep neural networks for fraud detection
- Adaptive edge weighting for graph-based learning algorithms
-
Structural-RNN: Deep Learning on Spatio-Temporal Graphs
-
greycat.ai
-
learning spatio temporal varying data
- Predictive Toxocology
- Computer Network Analysis / Management
- Resouce Allocation
- Fault tolerance
- Security
- Performance Monitoring
- Social Network Analysis
- Fraud Detection
- Transactional Data Analysis
- Supply Chain Optimisation
- Common tasks in signal-processing on graphs include filtering, de-noising, in-painting, compression, clustering, partitioning, sparsification, features extraction, classification and regression.
- SPMF - Data Mining Library