Masahiko Tateishi

Capabilities of a four-layered feedforward neural network: four layers versus three

Neural-network theorems state that only when there are infinitely many hidden units is a four-layered feedforward neural network equivalent to a three-layered feedforward neural network. In actual applications, however, the use of infinitely many hidden units is impractical. Therefore, studies should focus on the capabilities of a neural network with a finite number of hidden units, In this paper, a proof is given showing that a three-layered feedforward network with N-1 hidden units can give any N input-target relations exactly. Based on results of the proof, a four-layered network is constructed and is found to give any N input-target relations with a negligibly small error using only (N/2)+3 hidden units. This shows that a four-layered feedforward network is superior to a three-layered feedforward network in terms of the number of parameters needed for the training data.

Signal extraction system, system and method for speech restoration, learning method for neural network model, constructing method of neural network model, and signal processing system

A signal extraction system for extracting one or more signal components from an input signal including a plurality of signal components. This system is equipped with a neural network arithmetic section designed to process information through the use of a recurrent neural network. The neural network arithmetic section extracts one or more signal components, for example, a speech signal component and a noise signal component from an input signal including a plurality of signal components such as a speech and noises and outputs the extracted signal components. Owing to the presence of this neural network arithmetic section, the signal extraction becomes possible with a high accuracy.

Comments on "Artificial neural networks for four-coloring map problems and K-colorability problems"

The paper (see ibid., vol 38 p 326-33, 1991) gives a proof that the binary Hopfield neural network using the McCulloch-Pitts binary neuron model converges to the local minimum. In this correspondence the authors present a counter example to the proof and then point out the error in the proof. >

A Spoken Dialog Corpus for Car Telematics Services

Spoken corpora provide a critical resource for research, development and evaluation of spoken dialog systems. This chapter describes the spoken dialog corpus used in the design of CAMMIA (Conversational Agent for Multimedia Mobile Information Access), which employs a novel dialog management system that allows users to switch dialog tasks in a flexible manner. The corpus for car telematics services was collected from 137 male and 113 female speakers. The age distribution of speakers is balanced in the five age brackets of 20’s, 30’s, 40’s, 50’s, and 60’s. Analysis of the gathered dialogs reveals that the average number of dialog tasks per speaker was 8.1. The three most frequentlyrequested types of information in the corpus were traffic information, tourist attraction information, and restaurant information. Analysis of speaker utterances shows that the implied vocabulary size is approximately 5,000 words. The results are used for development and evaluation of automatic speech recognition (ASR) and dialog management software.

Robust Multimodal Dialog Management for Mobile Environments

This chapter describes three aspects of mobile dialog management: robustness in the presence of recognition errors; dynamic behavior based on user context (e.g. network connectivity, location); and efficient scenario description for multimodal dialogs. We describe algorithmic techniques for these three aspects of mobile dialog management, and results from empirical user studies are discussed which indicate significant improvement in performance and user satisfaction when these techniques are deployed in a dialog system.