https://github.com/TrustAGI-Lab/MTG

Joint Structure Feature Exploration and Regularization for Multi-Task Graph Classification
https://github.com/TrustAGI-Lab/MTG

Last synced: 3 months ago
JSON representation

Joint Structure Feature Exploration and Regularization for Multi-Task Graph Classification

• Host: GitHub
• URL: https://github.com/TrustAGI-Lab/MTG
• Owner: TrustAGI-Lab
• Created: 2015-09-01T12:37:27.000Z (about 9 years ago)
• Default Branch: master
• Last Pushed: 2022-02-28T09:24:23.000Z (over 2 years ago)
• Last Synced: 2024-07-15T16:49:54.700Z (4 months ago)
• Language: Java
• Homepage:
• Size: 9.59 MB
• Stars: 3
• Watchers: 1
• Forks: 9
• Open Issues: 0
• Metadata Files:
  • Readme: ReadMe.txt

Awesome Lists containing this project

• awesome-graph-classification - [Java Reference

README

        
MTG -- Source codes and Data used for 

	Shirui Pan, Jia Wu, Xingquan Zhu, Chengqi Zhang, and Philip Yu. Joint Structure Feature Exploration and Regularization for Multi-Task Graph Classification. TKDE, 2015.

Description: 

	This package includes two variants of MTG, i.e., MTG-l1 and MTG-l21. In general, MTG iteratively solves two subproblems: 

		(1) Multi-task learning (MTL) for vector data with logistic loss function

		(2) Most discriminative subgraph selection

	For the first subproblem, we employ MALSAR solver [1] to solve the multi-task problem. For the second subproblem, MTG  mploys a Top K subgraph miner in Java with upper bounds to prune the unpromising subgraph space.

Folders and Files:

	src/ : core scripts for MTG algorithm;

	MALSAR/ : a solver for solving the MTL problem;

	GMiner : Top-K discriminative subgrpah mining written in JAVA, it also provides source code for subgraph base graph classification, i.e., first mine a set of frequent subgraphs, and then employ SVMs for graph classification;

	data/ : NCI data used in the report

	mtg_result/ : results obtained from the demo

Demo:

	run demo_MTG.m for result

Other Reference

	1. J. Zhou, J. Chen and J. Ye. MALSAR: Multi-tAsk Learning via StructurAl Regularization. Arizona State University, 2012. http://www.public.asu.edu/~jye02/Software/MALSAR.

Tips:

If come across Out of Memory error, increase the Java Heap Space in Matlab:

    Preferences -> General -> Java Heap Space

    restart matlab