Embed Size (px)
DESCRIPTION
Diagnosis of Breast Cancer by Modular Evolutionary Neural Networks. Rahul Kala, School of Cybernetics, School of Systems Engineering University of Reading http://rkala.99k.org/ [email protected], [email protected]. - PowerPoint PPT Presentation
Citation preview
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Diagnosis of Breast Diagnosis of Breast Cancer by Modular Cancer by Modular Evolutionary Neural Evolutionary Neural NetworksNetworks
Rahul Kala,
School of Cybernetics, School of Systems Engineering
University of Reading
http://rkala.99k.org/
[email protected], [email protected] of paper: R. Kala, R. R. Janghel, R. Tiwari, A. Shukla (2011) Diagnosis of Breast Cancer by Modular Evolutionary Neural Networks, International Journal of Biomedical Engineering and Technology, 7(2): 194 – 211.
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Presentation to the paperPresentation to the paper
R. Kala et al. (2011) Diagnosis of Breast Cancer by Modular Evolutionary Neural Networks, International Journal of Biomedical Engineering and Technology [Accepted, In Press]
Research Sponsored by:
Indian Institute of Information Technology and Management Gwalior, INDIA
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Biomedical EngineeringBiomedical Engineering
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
The ProblemThe Problem
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Data SetData Set
Data Set Available At: W. H. Wolberg, O. L. Mangasarian, D. W. Aha. UCI Machine Learning Repository [http://www.ics.uci.edu/~mlearn/MLRepository.html], University of Wisconsin Hospitals, 1992.
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Machine Learning Machine Learning PerspectivePerspective
Data Set
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Not in agendaNot in agenda
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Concept – 1: Evolutionary Concept – 1: Evolutionary Neural NetworkNeural Network
Generate random individuals
While Stopping Criterion not met
Selection
Fitness Evaluation
Return best individual
Yes
No
Create Neural Network as per individual specifications
Training Algorithm as local search strategy
Performance/Connection Penalty Evaluation
Crossover
Mutation
Elite
Sys
tem
2
System 1
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Concept – 2: Attribute Concept – 2: Attribute DivisionDivision
Attribute Division
Inputs
Attribute Set 1
Attribute Set 2
Evolutionary Neural Net 1
Evolutionary Neural Net 2
Result Integration Output
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Concept – 3: Input Space Concept – 3: Input Space DivisionDivision
Input Space Clustering
Inputs
Cluster 1
Cluster 2
Attribute Division 1
Attribute Division 2
Result Integration Output
Cluster 3 Attribute Division 3
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Concept – 3: Input Space Concept – 3: Input Space DivisionDivision
Input Space Clustering
Inputs
Attribute Set 1
Attribute Set 2
Evolutionary Neural Net 1
Evolutionary Neural Net 2
Result Integration
Output
Cluster 1 Cluster 2 Cluster 3
Attribute Set 1
Attribute Set 2
Attribute Set 1
Attribute Set 2
Evolutionary Neural Net 2
Evolutionary Neural Net 2
Evolutionary Neural Net 2
Evolutionary Neural Net 2
Result Integration
Result Integration
Cluster Result Integration
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Concept 4: Mixture of Concept 4: Mixture of ExpertsExperts
Same input to all Experts
InputsInputs
Expert 1: Multi Layer Perceptron-1
Result Integration
OutputOutput
Expert 3: Multi Layer Perceptron-2
Expert 2: Radial Basis Function
Network
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
ResultsResultsS. No. Method Training
AccuracyTesting Accuracy
1. Proposed Algorithm 98.5075% 95.8084%
2. Modular Neural Network
94.4020% 91.4773%
3. Ensembles 98.2188% 94.8864%
4. Evolutionary Neural Network
96.2779% 95.7831%
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Component ResultsComponent ResultsCode Expert Number Cluster Number Module
NumberTraining Accuracy
Testing Accuracy
A Entire System (all experts combined) 98.5075% 95.8084%E1 Expert 1: Multi-Layer Perceptron (all clusters combined) 98.0100% 95.8084%E1.C1 Multi-Layer Perceptron-1 1 (all modules combined) 98.9362% 98.4848%E1.C1.M1 Multi-Layer Perceptron-1 1 1 98.9362% 95.4545%E1.C1.M2 Multi-Layer Perceptron-1 1 2 99.4681% 98.4848%E1.C2 Multi-Layer Perceptron-1 2 (all modules combined) 97.5000% 92.0635%E1.C2.M1 Multi-Layer Perceptron-1 2 1 93.3333% 84.1270%E1.C2.M2 Multi-Layer Perceptron-1 2 2 97.5000% 93.6508%E1.C3 Multi-Layer Perceptron-1 3 (all modules combined) 96.8085% 97.3684%E1.C1.M1 Multi-Layer Perceptron-1 3 1 96.8085% 97.3684%E1.C1.M2 Multi-Layer Perceptron-1 3 2 100% 96.8085% E2 Expert 2: Radial Basis Function Network 98.0100% 95.8084%E2.C1 Radial Basis Function 1 (all modules combined) 98.9362% 98.4848%E2.C1.M1 Radial Basis Function 1 1 97.8723% 93.9394%E2.C1.M2 Radial Basis Function 1 2 99.4681% 96.9697%E2.C2 Radial Basis Function 2 (all modules combined) 96.6667% 93.6508%E2.C2.M1 Radial Basis Function 2 1 94.1667% 88.8889%E2.C2.M2 Radial Basis Function 2 2 96.6667% 88.8889%E2.C3 Radial Basis Function 3 (all modules combined) 100% 97.3684%E2.C1.M1 Radial Basis Function 3 1 98.9362 97.3684%E2.C1.M2 Radial Basis Function 3 2 100% 97.3684%E3 Expert 1: Multi-Layer Perceptron (all clusters combined) 98.7562% 95.8084%E3.C1 Multi-Layer Perceptron-2 1 (all modules combined) 98.9362% 98.4848%E3.C1.M1 Multi-Layer Perceptron-2 1 1 98.9362% 95.4545%E3.C1.M2 Multi-Layer Perceptron-2 1 2 98.9362% 98.4848%E3.C2 Multi-Layer Perceptron-2 2 (all modules combined) 98.3333% 90.4762%E3.C2.M1 Multi-Layer Perceptron-2 2 1 95.0000% 85.7143%E3.C2.M2 Multi-Layer Perceptron-2 2 2 99.1667% 87.3016%E3.C3 Multi-Layer Perceptron-2 3 (all modules combined) 98.9362% 94.7368%E3.C1.M1 Multi-Layer Perceptron-2 3 1 100.0000% 97.3684%E3.C1.M2 Multi-Layer Perceptron-2 3 2 98.9362% 97.3684%
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Related Publications - Related Publications - JournalsJournals Kala, Rahul, Tiwari, Ritu, & Shukla, Anupam (2011) Breast Cancer
Diagnosis using Optimized Attribute Division in Modular Neural Networks, Journal of Information Technology Research, Vol. 4, No 1, pp 34-47
Kala, Rahul, Janghel, Rekh Ram, Tiwari, Ritu, & Shukla, Anupam, (2011) Diagnosis of Breast Cancer by Modular Evolutionary Neural Networks, International Journal of Biomedical Engineering and Technology, Inderscience [In Press]
Kala, Rahul, Vazirani, Harsh, Khawalkar, Nishant, & Bhattacharya, Mahua (2010) Evolutionary Radial Basis Function Network for Classificatory Problems, International Journal of Computer Science Applications, TMRF India, Vol. 7, No. 4, pp 34-49
Kala, Rahul, Vazirani, Harsh, Shukla, Anupam, & Tiwari, Ritu (2010) Medical Diagnosis using Incremental Evolution of Neural Network, Journal of Hybrid Computing Research, TMRF India , Vol. 3, No. 1, pp 9-17
Kala, Rahul, Vazirani, Harsh, Shukla, Anupam, & Tiwari, Ritu (2010) Evolution of Modular Neural Network in Medical Diagnosis, International Journal of Applied Artificial Intelligence in Engineering System, Vol. 2, No. 1, pp 49 -58
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Related Publications - Related Publications - ConferencesConferences Meena, Yogesh Kumar, Arya, Karam Veer, Kala, Rahul (2010) Classification using
Redundant Mapping in Modular Neural Networks, Proceedings of the 2010 World Congress on Nature and Biologically Inspired Computing, Kitakyushu, Japan [In Press]
Janghel, R. R., Shukla, Anupam, Tiwari, Ritu, Kala, Rahul (2010) Breast Cancer Diagnostic System using Symbiotic Adaptive Neuro-evolution (SANE). Proceedings of the 2010 International Conference of Soft Computing and Pattern Recognition, Cercy Pontoise/Paris, France, pp 326-329.
Janghel, R. R., Shukla, Anupam, Tiwari, Ritu, Kala, Rahul (2010) Breast Cancer Diagnosis using Artificial Neural Network Models. Proceedings of the IEEE 3rd International Conference on Information Sciences and Interaction Sciences, pp 89-94, Chengdu, China.
Vazirani, Harsh, Kala, Rahul, Shukla, Anupam, Tiwari, Ritu (2010) Diagnosis of Breast Cancer by Modular Neural Network. Proceedings of the Third IEEE International Conference on Computer Science and Information Technology, pp 115-119, Chengdu, China
Kala, Rahul, Shukla, Anupam, & Tiwari, Ritu (2009) Comparative analysis of intelligent hybrid systems for detection of PIMA indian diabetes, Proceedings of the IEEE 2009 World Congress on Nature & Biologically Inspired Computing, NABIC '09, pp 947 - 952, Coimbatote, India
Kala, Rahul, Shukla, Anupam, Tiwari, Ritu (2009) Fuzzy Neuro Systems for Machine Learning for Large Data Sets, Proceedings of the IEEE International Advance Computing Conference, IACC '09, pp 541-545, Patiala, India
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
More from the authorsMore from the authors
School of Cybernetics, School of Systems Engineering,
University of ReadingPresentation Skills WorkshopMarch 22, ‘11
Thank You Thank You Rahul Kala
Call Centre Lab, Room No 188,
School of Cybernetics,
School of Systems Engineering,
University of Reading, Whiteknights
http://rkala.99k.org/
Ph: +44 (0) 7424752843
Acknowledgements
Prof. Anupam Shukla,
Professor, IIITM Gwalior
Dr. Ritu Tiwari,
Assistant Professor, IIITM Gwalior
Mr. R. R. Janghel,
Research Scholar, IIITM Gwalior
