Michael Madden IT Department NUI Galway
Home
Software for Bayesian Network Classification

Over several years, I have developed open-source machine learning software for inductively learning various forms of Bayesian networks from data, and for classification using BNs.

You can download it here, with instructions on how to use it.

Features include:
  • Learn Naive Bayes, Tree-Augmented Naive Bayes (TAN), General Bayesian Networks, and BN-Augmented Naive Bayes structures
  • Multiple scoring metrics, including the K2 metric, BDeu and CMI
  • Visualisation of structures using GraphViz (DOT and DOTTY)
  • Supplementary code to work with data from other formats, to perform cross-validation, to construct learning curves, and to construct ROC curves
For a description of the classifiers, and a discussion of their relative performance, see: “On the Classification Performance of TAN and General Bayesian Networks” , Michael G. Madden. Knowledge Based Systems, 2009.

Please cite that paper if you use this software in your work.

 
New PhD scholarships - Sept 2014

September 2014:

We have 1 or 2 extra PhD scholarships, including a stipend of up to €13,000 tax free per annum + fees + some travel money, available for highly qualified applicants. 

If you are interested in working with me in the Data Mining and Machine Learning Research Group, please contact me by email for details.

Applicants should be available to start by October 2014. Applicants should have:

  • A BSc or MSc in Computer Science or a closely related discipline, with a track record of scholarship that indicates excellent potential for a PhD research
  • Excellent skills in communicating and helping others learn, as these activities will be required as part of the scholarship conditions
  • Evidence of strong programming abilities and comfort with mathematical concepts.

Your research can be in any topic related to my research group's interests. This link has some research topic ideas.

For these scholarships, there will be an evaluation process involving shortlisting based on your background, an interview, and an assessment of your ability to tutor students in multiple programming languages. 

Deadline: Contact me as soon as possible. Applications must be received by mid September.

 
Research Overview

The Data Mining and Machine Learning Group is led by Michael Madden of the College of Engineering & Informatics, NUI Galway.

Our research is focused on new theoretical advances in machine learning and data mining, motivated by important practical applications, on the basis that challenging applications foster novel algorithms which in turn enable new applications.

Specific research topics include:

  • Artificial intelligence, data mining & machine learning
  • Algorithms for classification and numeric prediction
  • New methods for combining domain knowledge with data mining
  • Time series data analysis
  • Probability, reasoning under uncertainty, and Bayesian networks
  • Reinforcement learning
  • Practical applications of data mining and machine learning in science, engineering & medicine.

For more specific information, take a look at our current and recent research projects, our publications, and the current and past members of our group. You can also contact me with questions.

 
Nonlinearity and Uncertainty in Drug Modelling

Project Overview

This is a strongly interdisciplinary collaborative project, involving Informatics, Mathematics and Applied Mathematics, with an application in Medical Informatics. The project collaborators include researchers in Mathematics and Applied Mathematics in NUI Galway, the Intensive Care Unit at University Hospital Galway, as well as Prof. Stuart Russell’s group in Computer Science in the University of California at Berkeley.

In an ICU ward, computer systems constantly monitor and record data such as changes in heart rate and blood pressure. This project will develop software techniques that uses this data, and will add new mathematics and computing techniques that infer patient response to certain drugs in real-time, to predict the impact on the patient of different drug regimes. This will allow for drug treatments to be tailored to the individual. This contrasts with the current practice of basing treatments on standardised data, taken from clinical trials using healthy volunteers, who might not respond in the same way as sick patients. 

Read more...
 

© M Madden