Fusion of External Context and Patterns – Learning from Video Streams
A mathematical model, which extends the Bayesian problem of pattern recognition by fusion of external context variables and patterns is proposed and investigated. Then, its empirical version is discussed and a learning algorithm for an orthogonal neural net is proposed, which takes context variables into account. The proposed algorithm has a recursive form, which is well suited for learning from a stream of patterns, which arise when features are extracted from a video sequence.
KeywordsContext Variable Learning Sequence Smoke Alarm Recursive Form Pattern Recognition Problem
Unable to display preview. Download preview PDF.
- 3.Greblicki, W.: Asymptotically Optimal Pattern Recognition Procedures with Density Estimates. IEEE Trans. Information Theory IT 24, 250–251 (1978)Google Scholar
- 4.Greblicki, W.: Asymptotic Efficiency of Classifying Procedures using the Hermite Series Estimate of Multivariate Probability Densities. IEEE Trans. Information Theory IT 17, 364–366 (1981)Google Scholar
- 6.Rafajłowicz, E.: Context-dependent neural nets – Problem statement and examples. In: Proc. 3rd Conf. Neural Networks Applications, Zakopane, Poland, May 1999, pp. 557–562 (1999)Google Scholar
- 7.Rafajłowicz, E.: Learning context dependent neural nets. In: Proc. 3rd Conf. Neural Networks Applications, Zakopane, Poland, May 1999, pp. 551–556 (1999)Google Scholar
- 8.Ruta, D., Gabrys, B.: An Overview of Classifier Fusion Methods. Computing and Information Systems 7, 1–10 (2000)Google Scholar
- 10.Turney, P.: The identification of context-sensitive features: A formal definition of context for concept learning. In: Proc. 13th Int. Conf. Machine Learning (ICML 1996), Bari, Italy, pp. 53–59 (1996)Google Scholar