12 March 2002 Self-similarity for data mining and predictive modeling: a case study for network data
Author Affiliations +
Recently there are a handful study and research on observing self-similarity and fractals in natural structures and scientific database such as traffic data from networks. However, there are few works on employing such information for predictive modeling, data mining and knowledge discovery. In this paper we study, analyze our experiments and observation of self-similar structure embedded in Network data for prediction through Self Similar Layered Hidden Markov Model (SSLHMM). SSLHMM is a novel alternative of Hidden Markov Models (HMM) which proven to be useful in a variety of real world applications. SSLHMM leverage HMM power and extend such capability to self-similar structures and exploit this property to reduce the complexity of predictive modeling process. We show that SSLHMM approach can captures self-similar information and provides more accurate and interpretable model comparing to conventional techniques and one-step, flat method for model constructions such as HMM.
© (2002) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Jafar Adibi, Jafar Adibi, Wei-Min Shen, Wei-Min Shen, Eaman Noorbakhsh, Eaman Noorbakhsh, } "Self-similarity for data mining and predictive modeling: a case study for network data", Proc. SPIE 4730, Data Mining and Knowledge Discovery: Theory, Tools, and Technology IV, (12 March 2002); doi: 10.1117/12.460227; https://doi.org/10.1117/12.460227

Back to Top