Noninformative priors for linear function of parameters in the lognormal distribution

Abstract

This paper considers the noninformative priors for the linear function of parameters in the lognormal distribution. The lognormal distribution is applied in the various areas, such as occupational health research, environmental science, monetary units, etc. The linear function of parameters in the lognormal distribution includes the expectation, median and mode of the lognormal distribution. Thus we derive the probability matching priors and the reference priors for the linear function of parameters. Then we reveal that the derived reference priors do not satisfy a ?rst order matching criterion. Under the general priors including the derived noninformative priors, we check the proper condition of the posterior distribution. Some numerical study under the developed priors is performed and real examples are illustrated.

English Speech Recognition and Evaluation of Pronunciation Quality Using Deep Learning

“…In the twenty-first century, especially since 2009, the development of technology regarding recognition of speech has accelerated which is directly related to the rapid development of recognition of speech technology using deep learning, and accumulates a huge quantity of original speech data at the same time. Lee and other scholars take the modified linear function as the excitation function of hidden layer nodes, and use recursive neural network [21] in music processing. Li Deng et al Recognized different languages and phonemes on the basis of multi-layer result conditional random fields [22].…”