Neural networks are a useful alternative to Gaussian mixture models for acoustic modeling; however, training multilayer networks involves a difficult, nonconvex optimization that...
: Kriging-based exploration strategies often rely on a single Ordinary Kriging model which parametric covariance kernel is selected a priori or on the basis of an initial data set....
In this paper we propose a Gaussian-kernel-based online kernel density estimation which can be used for applications of online probability density estimation and online learning. ...
This paper proposes a background subtraction method for Bayer-pattern image sequences. The proposed method models the background in a Bayer-pattern domain using a mixture of Gauss...
Abstract. A sign language recognition system based on Hidden Markov Models(HMMs) and Auto-regressive Hidden Markov Models(ARHMMs) has been proposed in this paper. ARHMMs fully cons...
Xiaolin Yang, Feng Jiang, Han Liu, Hongxun Yao, We...