In speech recognition systems, a common problem is transcription of new additions to the recognition lexicon into their phonetic symbols. Specific to the Japanese language, such a problem can be dealt with in two steps. We focus on the first step, in which the new lexical entry is converted into a set of hiragana syllabaries, which is almost a phonetic transcription. We propose a conversion scheme which yields the most likely hiragana syllabaries, based on a language model. Results from our evaluations on three test sets are also reported. Although the study is conducted on Japanese only, our approach has applications to Chinese.

Nonparametric estimation of interaction functions for two-type pairwise interaction point processes is addressed. Such a problem is known to be challenging due to the intractable normalizing constant present in the density function. It is shown that the means of the marked interpoint distance functions embedded in the two-type pairwise interaction point process converge to the means of an inhomogeneous Poisson process. This suggests a simple and effective nonparametric estimation method. An example is presented to illustrate the efficacy of our method. Our results can be generalized to multitype point processes in a straightforward manner, although the notation is more involved.

The sufficient statistic for performing the likelihood ratio test for pairwise interaction point processes is well-known; however, the evaluation of its performance is a very difficult problem. It is shown that the distribution of the sufficient statistic can be approximated by the distribution of a Poisson-driven shot-noise random variable, which can be readily computed.