Incremental clustering techniques
Recent years have witnessed the explosive growth of online data. Unlike traditional offline data, online data has its unique characteristics: constantly evolving and arriving in streaming manner. Many online clustering methods have been proposed to efficiently handle the online data. In this project...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/54623 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Recent years have witnessed the explosive growth of online data. Unlike traditional offline data, online data has its unique characteristics: constantly evolving and arriving in streaming manner. Many online clustering methods have been proposed to efficiently handle the online data. In this project, the incremental Spectral Clustering (iSC) algorithm [1] has been researched and implemented. The iSC algorithm can efficiently handle the changes, insertions or deletions of data objects by incrementally updating eigenvalue system. Additionally, some iSC related topics have been explored and implemented, which includes the data grouping technique, the automatic determination of the number of clusters and the clustering result matching. Moreover, this project also studied and implemented the online Non-negative Matrix Factorization (NMF) algorithm [2] to gain more exposure in the field of clustering. Afterwards, various experiments have been conducted to evaluate the above mentioned algorithms and techniques. |
|---|