Hanan Ayad Supervisor Prof. Mohamed Kamel

1. Developing Methods for Combining Multiple Clustering of PatternsTowards the discovery of natural clusters Hanan Ayad Supervisor Prof. Mohamed Kamel

2. Outline • Motivations • Research Summary • 2003 Publications • Application to Learning Objects • Diverse Sources of Information • Multiple Clustering in LO • Process Overview

3. Motivations • Multiple clustering solutions • Different clustering methods • Selection of learning parameters (e.g. NNets) • Random starts, random ordering of patterns • Enhance Quality • Compensatory effects in clustering methods • Repeated fine decompositions • Distributed clustering • Feature-Distributed • Multiple partial views • Random subspaces • Alternative feature reductions • Data-Distributed • Overlapping subsets of patterns

4. Research Summary • Measure co-associations between patterns based on their co-clustering - voting • Development of combination rules based on shared co-associations – Shared nearestneighbors(binary votes, weighted votes, sum rule, product rule, rank-based rule) • Determine strength-of-association • Accumulate local neighborhood densities of patterns • Patterns weights are inversely proportional to their local neighbourhood densities(number and weights of relationships)

5. Research Summary, Cont’d • Pruning of associations for efficiency and assessment of behaviour. • Effect on mutuality of relationships • Improving quality using subsets of patterns relations • Study of convergence and stability • Induce a graph, representing the patterns weighted relationships, and the patterns own weights. Weighted Shared nearest neighbors Graph (WSnnG) • The graph is partitioned resulting in an integrated clustering of the objects • Use of the graph partitioning package METIS. • Minimize edge-cut subject to weights of vertices being equally distributed among the clusters

6. 2003 Publications • H. Ayad and M. Kamel. "Finding Natural Clusters Using Multi-Clusterer Combiner Based on Shared Nearest Neighbors“. Multiple Classifier Systems: Fourth International Workshop, MCS 2003, Guildford, Surrey, United Kingdom, June 11-13. Proceedings. • H. Ayad, and M. Kamel. Refined Shared Nearest Neighbors Graph for Combining Multiple Data Clusterings", The 5th International Symposium on Intelligent Data Analysis IDA 2003. Berlin, Germany. Proceedings. LNCS. Springer. August, 2003 • H. Ayad, and M. Kamel. Development of New Methods for Combining Cluster Ensembles. On going Journal Paper.

7. Application to Learning Objects Diverse Sources of Information • Meta Data • Standardized Indexing, content structure and organization • Intelligent Content Mining • Natural Language Understanding • Image Analysis and Understanding • Automatic Speech Recognition • Statistical Learning • Re-Use/Learning Scenarios • Dynamic assembly, object are grouped and regrouped with other objects.

8. Application to Learning Objects Multiple Clustering in LO • Clusters of Learning Objects • Multiple distributed taxonomies • Info. Sources: different sets of meta data, different re-use/assembling scenarios • Dynamic environment, clusters based on partial views • Combining of multiple clustering • Mining complex web of relationships • integrating multiple objects clustering • Discovery of combined multi-view clusters.

9. Meta Data 1 . . . Meta Data m Re-use Scenarios 1 . . . Re-use Scenarios s Application to Learning Objects Process Overview Clustering 1 . . . Combining Multiple Objects Clustering Integrated Learning Objects Clusters Learning Objects Clustering r