University of Nottingham Malaysia
School of Computer Science
     
  

Machine Learning

Overview

This research aims to understand and contribute to both pattern recognition and machine learning in order to predict an outcome and/or decide on a suitable course of action. In all these cases the inference engine used is the Support Vector Machine, a linear classifier with good generalization capability which is due to its approximate implementation of structural risk minimization. Overview 

* All e-mails are @nottingham.edu.my

Current projects

Machine learning projects
Proj1  

Title. System for Super-capacitor Pilot plant production.
Description. This project deals with producing a prototype system (software) which can be installed in a super-capacitor factory to aid the manufacturing personnel in solving production problems as efficiently and as quickly as possible. The objective of this project is to design and implement software which facilitates the storage and intelligent retrieval of expert knowledge from an organized knowledge base in the manufacturing environment based on specific user characteristics. The system should also ultimately be able to predict user need based on his or her profile thru the classification and clustering of user characteristics which are closely related to each other.

 Proj2

Title. Monitoring and Prediction of Blood Sugar level for Diabetics
Description. The objective of our study is to design and implement a User Modelling System for the diabetic patient to determine and predict blood glucose level through the use of specially designed software. With this system, blood glucose readings are taken only once every two days as the software is able to track changes in blood sugar depending on the food eaten and other relevant factors over that period. A database of response to common foods is set up and background information of the user is used in conjunction with his past response to predict his future response to new food and situations he may encounter thus eliminating the need for multiple daily blood glucose tests.

ChrisProj1

Title. Ensemble Learning for Colorectal Cancer Prognosis
Description. This project deals the use of several disparate modelling techniques on a highly dimension cancer dataset. By using subsets of the data and methods such as anti-learning, class leading predictions can be made for survival rates of certain cancer patients. Clustering, semi-supervised and supervised learning are all used to better understand the data and select the appropriate data subsets. Once accurate ensemble learning models are generated, model transparency methods can be used to highlight key cause effect relationships within the data and these finding can be used by clinicians to better treat patients.

Proj3  

Title: Detection and Prediction of Lung Cancer use the zNose with the Support Vector Machine Classifier
Description. We propose a Lung Cancer Detection system which is a hybrid breath test and patient models incorporates Case-Based Reasoning Cycle system to help multivariate analysis information in order to make diagnose decisions more inexpensively accurately and rapidly. The main issue of implementing this hybrid system is a lung cancer knowledge base which is derived from the National Cancer Patient Registry the first database to record detailed information on cancer patients in Malaysia. By using artificial intelligent tool support vector machine to serve as data mining engine, the lung cancer patients’ characteristic such as gender, age, smoking history etc. is founded; the patient model was made up by fusing these characters and the breath test sample (volatile organic compounds) which is collected from a portable breath collection apparatus zNose. Certain number of healthy volunteers will be imported as another class for the knowledge base in order to accomplish prediction function; two prediction values is determined in the case-based reasoning cycle, one for lung cancer patient model and zero for no disease.  The objective of this project is to combine data mining technology and artificial intelligence classifiers as a means to construct lung cancer patient models and to link this to the case-based reasoning cycle in order to provide precisely diagnose of lung cancer in a timely manner.

Funding

Grant: Ministry of Science, Technology and Innovation, Malaysia for the first level of Technofunds.
Title: Super-capacitor Pilot Plant project
Dates: from 2008 to 2010.

Grant: Ministry of Science, Technology and Innovation, Malaysia for eScience.
Title: Detection and Prediction of Lung Cancer use the zNose with the Support Vector Machine Classifier
Dates: from 2010 to 2011.

Selected publications

Journal Papers:

  1. Chen ZhiYuan, Dino Isa, Peter Blanchfield and Timothy Brailsford, "Monitoring and Predicting of Blood Sugar Level by Using HDCU", submitted to IEEE Transactions on Knowledge and Data Engineering, 2011.
  2. Chen ZhiYuan, Dino Isa, Peter Blanchfield and Roselina Arelhi, "Study on the Correlation Coefficient of Gene Expression within Normal Lung and Carcinoid Class", submitted to IEEE Transactions on Information Technology in Biomedicine, 2011.
  3. Chen ZhiYuan, Dino Isa and Peter Blanchfield, “Improve the Classification and Prediction Performance for the IP Management System in a Super-capacitor Pilot Plant", International journal of Latest Trends in Computing (IJLTC), Volume 1 Issue 2, December 2010.
  4. Chris Roadknight, Uwe Aickelin, Galina Sherman, Validation of a Microsimulation of the Port of Dover, Journal of Computational Science, Available online 28 July 2011, ISSN 1877-7503, 10.1016/j.jocs.2011.07.005
  5. Dino Isa, Chen ZhiYuan and Peter Blanchfield, "Vectorization Algorithm for an Intelligentized System”. International Journal of Computer and Network Security, Volume 2 February issue, No2, 2010.

Conference Papers:

  1. Chen ZhiYuan, Dino Isa and Peter Blanchfield, " An Artificial Intelligence Methodology  for Intellectual Property Management", 16th International Conference on Neural Information Processing, Workshop on Advances in Intelligent Computing, Kuala Lumpur, Malaysia 2009
  2. Chen ZhiYuan, Dino Isa and Peter Blanchfield, “Intellectual Property Management system for the Super-capacitor Pilot Plant", International Conference on Artificial Intelligence, Las Vegas, USA 2009.
  3. Chen ZhiYuan, Dino Isa and Peter Blanchfield, "A Hybrid Data Mining and Case-Based Reasoning User Modelling System (HDCU) for Monitoring and Predicting of Blood Sugar Level", International Conference on Computer Science and Software Engineering, WuHan, China 2008.
  4. Chen ZhiYuan, Dino Isa and Peter Blanchfield, "Column Vectorizing Algorithms for Support Vector Machines", World Congress on Engineering & Computer Science, International Conference on Machine Learning and Data Analysis, San Francisco, USA 2008.
  5. Chen ZhiYuan, Dino Isa and Peter Blanchfield, "A Hybrid Data Mining and Case-Based Reasoning User Modelling System Architecture", World Congress on Engineering 2008, International Conference of Computational Intelligence and Intelligent Systems, London, UK 2008.
  6. Chris Roadknight, Uwe Aickelin, Alex Ladas, Daniele Soria, John Scholefield and Lindy Durrant  Biomarker Clustering of Colorectal Cancer Data to Complement Clinical Classification. FEDERATED CONFERENCE ON COMPUTER SCIENCE AND INFORMATION SYSTEMS.  2012
  7. Chris Roadknight, Uwe Aickelin, Guoping Qiu, John Scholefield and Lindy Durrant.  Supervised Learning and Anti-learning of Colorectal Cancer Classes and Survival Rates from Cellular Biology Parameters. IEEE International Conference on Systems, Man, and Cybernetics 2012
  8. Dino Isa Chen ZhiYuan and Peter Blanchfield, "Data pre-processing in a Hybrid Data Mining and Case-Based Reasoning User Modelling System", the 3rd Malaysian Software Engineering Conference, Kuala Lumpur, Malaysia 2007.

Main contacts

  • Chen ZhiYuan (web page, e-mail: Zhiyuan.Chen*).
  • Chris Roadnight (web page, e-mail: Chris.Roadknight*). 

Links

* e-mail suffix: @nottingham.edu.my

School of Computer Science

The University of Nottingham Malaysia Campus
Jalan Broga, 43500 Semenyih
Selangor Darul Ehsan
Malaysia

telephone: +6 (03) 8924 8767
fax: +6 (03) 8924 8018

Make an enquiry