Applied Stochastic Models and Data Analysis

1990 Applied Stochastic Models and Data Analysis

doi: 10.1002/asm.3150060301

Boyer, A.; Di Martino, J.; Divoux, P.; Haton, J. P.; Mari, J. F.; Smaili, K.

1990 Applied Stochastic Models and Data Analysis

doi: 10.1002/asm.3150060302

Automatic speech recognition and understanding (ASR) plays an important role in the framework of man‐machine communication. Substantial industrial developments are at present in progress in this area. However, after 40 years or so of efforts several fundamental questions remain open. This paper is concerned with a comparative study of four different methods for multi‐speaker word recognition: (i) clustering of acoustic templates, (ii) comparison with a finite state automaton, (iii) dynamic programming and vector quantization, (iv) stochastic Markov sources. In order to make things comparable, the four methods were tested with the same material made up of the ten digits (0 to 9) pronounced four times by 60 different speakers (30 males and 30 females). We will distinguish in our experiments between multi‐speaker systems (capable of recognizing words pronounced by speakers that have been used during the training phase of the system) and speaker‐independent systems (capable of recognizing words pronounced by speakers totally unknown to the system). Half of the corpus (15 male and 15 female) were used for training, and the remaining part for test.

Owsiński, Jan W.

1990 Applied Stochastic Models and Data Analysis

doi: 10.1002/asm.3150060303

A new clustering method is presented which proposes a class of objective functions and an algorithm which sub‐optimizes the objective functions over the whole space of partitions. The objective functions have a global nature, encompassing both the cluster contents and the cluster number. However, the accompanying suboptimization algorithm works according to a simple progressive merger scheme. The algorithmic scheme produces in a quite natural way an indexed hierarchy. The hierarchy index is not just tacked on to the method—see Diday and Moreau1—on the contrary, the algorithm refers directly to its values which measure, depending upon the particular formulation, either the relative affinity or the relative difference of the two clusters merged at a given level of hierarchy. In this way, the scale of hierarchy and hierarchy‐wise validity of clusters can easily be established, which is of great importance in analysing unstructured data sets whose generating process is unknown and can only be hypothesized after an initial structure had been established, e.g. owing to clustering, as is the case in pattern recognition—see Kaminuma2.

Tralhǎo, Lino M.; Craveirinha, José F.; Clímaco, Joáo N.

1990 Applied Stochastic Models and Data Analysis

doi: 10.1002/asm.3150060304

In the first part of this study we present and review a simplified model for the traffic flow between the switches of a modular switching system. In the second part, an approximated model for calculating the blocking probability on a system with this type of architecture is presented and then generalized to a structure here defined as an m‐asymmetrical incomplete link system. The model is based on the probabilistic hypothesis of Jacobaeus and leads to a system of formulae which may be calculated using a computing language allowing for recursive subprograms.

Troussart, Jacques

1990 Applied Stochastic Models and Data Analysis

doi: 10.1002/asm.3150060305

1990 Applied Stochastic Models and Data Analysis

doi: 10.1002/asm.3150060306