Information Technology Reference
In-Depth Information
Thus, when the number of the clusters is small, the performance of the NNSC
and NMSC is almost same. However, in practice, there always exist much more
clusters in the data set, especially in the document corpora. From the perfor-
mance shown in the Figure 3, the proposed NMSC proves more acceptable than
the other methods in processing the multiple-cluster dataset.
4 Conclusions
In this paper, we have put forward non-negative mutative-sparseness coding as a
useful framework to explore hidden components and hierarchical representation
of data. Uncommonly, through constantly adjusting sparseness of the data in the
process of further analyzing respective bases, all the data can be well described
and grouped into proper clusters of multiple layers. In the experimental evalu-
ations, NMSC embodies its great eciency in clustering and sucient merit in
hierarchical representing documents. Then for practical application, NMSC will
help to boost the development of patent system, library system, etc. In addition,
towards future work, we also aim to test NMSC in other fields such as image
and audio processing.
References
1. Dumais, S.T.: Latent Semantic Analysis. Annual Review of Information Science
and Technology 38, 188 (2005)
2. Hofmann, T.: Probabilistic Latent Semantic Indexing. In: Proceedings of the
Twenty-Second Annual International SIGIR Conference on Research and Develop-
ment in Information Retrieval, SIGIR 1999 (1999)
3. Blei, D.M., Ng, A.Y., Jordan, M.I.: Latent Dirichlet Allocation. Journal of Machine
Learning Research 3, 993-1022 (2003)
4. Lee, D.D., Sebastian Seung, H.: Algorithms for Non-negative Matrix Factorization,
pp. 556-562. MIT Press (2001)
5. Hoyer, P.O.: Non-negative sparse coding. In: Proceedings of the 12th IEEE Work-
shop, pp. 557-565 (2002)
6. Ghahramani, Z.: Graphical models (Speech). Tubingen, Germany (2008) (re-
trieved)
7. Buntine, W.L.: Operations for Learning with Graphical Models. Journal of Artifi-
cial Intelligence Research (AI Access Foundation) 2, 159-225 (1994) ISSN: 11076-
9757
8. Yuan, X.: A Clustering Algorithm for Web Document Based on Theme. Journal
of Chengdu University (Natural Science Edition) 29(3) (2010)
9. Han, K.: Extract Subject from Chinese Text with Three Different Levels. Journal
of Chinese Information 15(4) (2001)
10. Zha, H., Ding, C., Gu, M., He, X., Simon, H.: Spectral relaxation for k-means
clustering. In: Advances in Neural Information Processing Systems, vol. 14 (2002)
11. Shi, J., Malik, J.: Normalized cuts and image segmentation. IEEE Transactions on
Pattern Analysis and Machine Intelligence 22(8), 888-905 (2000)
Search WWH ::
Custom Search