ECE 8820 Pattern Recognition               

Computer Project #3                        

Fall, 2008

The purpose of this project is to experiment with different clustering algorithms and to compare the results. 

Part I: Design and implement the hard C-Means algorithm (HCM) and the fuzzy C-Means algorithm (FCM). Experiment with different distance measures and/or different values for the fuzzifier. Compare the HCM and FCM results.

Part II: Design and implement an agglomerate hierarchical clustering algorithm (try single link and complete link), and compare the results to the HCM and FCM algorithms.

Several data sets are provided in the form of 2D PGM image files.  You can download a zip file with the PGM files and image utilities written in C. Alternatively, you may also use your own dataset.

As before, your report should contain sections on

• The technical description of all techniques utilized,
• The design of the algorithms (pseudo-code, flowcharts, or some other structured descriptive means),
• The results of the algorithms (Describe your experimental conditions with the results so it is clear what experiment was done for each set of results.)
• An analysis of the results, i.e., did you obtain what you expected?  Were there any surprises? What conclusions can you draw from the experiments? etc.  
• Well documented, structured, modular program listings.