• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Real-time speech recognition and model selection via recurrent neural networks

Research Project

Project/Area Number 06650401
Research Category

Grant-in-Aid for General Scientific Research (C)

Allocation TypeSingle-year Grants
Research Field 情報通信工学
Research InstitutionThe University of Electro-Communications

Principal Investigator

TAKAHASHI Haruhisa  The Univ.of Electro-Communications, Dept.of Communications and Systems Eng., Associate Prof., 電気通信学部, 助教授 (90135418)

Co-Investigator(Kenkyū-buntansha) YOSHIDA Toshinobu  The Univ.of Electro-Communications, Dept of Computer Sciences and Information Ma, 電気通信学部, 助教授 (30114341)
TOMITA Etsuji  The Univ.of Electro-Communications, Dept.of Communications and Systems Eng., Pro, 電気通信学部, 教授 (40016598)
Project Period (FY) 1994 – 1995
Project Status Completed (Fiscal Year 1995)
Budget Amount *help
¥2,000,000 (Direct Cost: ¥2,000,000)
Fiscal Year 1995: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 1994: ¥1,500,000 (Direct Cost: ¥1,500,000)
KeywordsSpeech recognition / Neural networks / Machine learning / Probably Approximately Correct
Research Abstract

We performed the study on the theme of this report by intensively investigating the theoretical base of learning. In the first year we developed a very simple recurrent neural network (VSRN) architecture which is a three-layr network and contains only self-loop recurrent connections in the hidden layr. The role of the recurrent connection is explained by the network dynamics and its function will be acquired by learning from finite examples like a mamalian action. Through the learning process some characteristic functions observed in the mamalian auditory systems are founed automatically acquired by the network. In the second year we investigated mainly the theoretical framework of how our network can learn well by proposing a new method for analysing the generalization performance. To achieve this, we undertake a comparison of learning and hypothesis testing, which leads to a novel notion of regular interpolation dimension and an ill-disposed learning algorithm that produces ill-disposed hypotheses. This unites the learning and the hypothesis testing in a common viewpoint such that the base of hypothesis testing inequalities can be directly used for estimating ill-disposed hypotheses on training examples. The regular interpolation dimension is no greater than the number of modifiable system parameters. We analyze the ill-disposed learning algorithm both in the PAC learning model and in an average-case setting to obtain more explicit bounds on learning curves and sample complexity in terms of the regular interpolation dimension, than those in terms of the VC dimension. The results are applied and extended to the other algorithm such as a Gibbs algorithm and the inconsistent learning to obtain explicit bound of the learning curves and sample complexity.

Report

(3 results)
  • 1995 Annual Research Report   Final Research Report Summary
  • 1994 Annual Research Report
  • Research Products

    (18 results)

All Other

All Publications (18 results)

  • [Publications] Takahashi H.: "Towards practical bounds of storage capacity in multilayer networks." Proceedings of NOLTA'95. 2B-5. 215-218 (1995)

    • Description
      「研究成果報告書概要(和文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Gu Hanzhong: "Self-Averaging and Sample Complexity." 信学技報. NC96-1. 1-8 (1996)

    • Description
      「研究成果報告書概要(和文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Gu Hanzhong: "Exponential or polynomial learning curves? a case study" Proceedings of NOLTA'95. 2B-12. 243-246 (1995)

    • Description
      「研究成果報告書概要(和文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] 顧漢忠: "最悪学習曲線の上限-実用的評価を目指して" 信学技報. NC95-67. 15-22 (1995)

    • Description
      「研究成果報告書概要(和文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] 顧漢忠: "概念学習における学習曲線の評価." 信学技報. NC95-57. 63-70 (1995)

    • Description
      「研究成果報告書概要(和文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Gu Hanzhong: "Towards More Practical Average Bounds on Supervised Learning." IEEE Transactions on Neural Networks. 21 (1996)

    • Description
      「研究成果報告書概要(和文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Takahasi H.: "Statistical epsilon-Capacity of Multilayr Networks and eneralization Performance." Proceedings of Symposium on NOLTA. 271-274 (1994)

    • Description
      「研究成果報告書概要(欧文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Takahashi H.: "Towards practical bounds of storage capacity in multilayr networks." Proceedings of Simposium on NOLTA. 2B-5. 215-218 (1995)

    • Description
      「研究成果報告書概要(欧文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Gu H,Takahashi H.: "Self-Averaging and Sample Complexity." Technical Report of IEICE. NC-95-67. (1996)

    • Description
      「研究成果報告書概要(欧文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Gu H,Takahashi H.: "Exponential or polynomial learning curves? a case study"." Proceedings of Simposium on NOLTA. 2B-12. 243-246 (1995)

    • Description
      「研究成果報告書概要(欧文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Gu H,Takahashi H.: "Generalizations of the PAC model for function estimation." Proceedings of Simposium on NOLTA,October. 6-8 (1994)

    • Description
      「研究成果報告書概要(欧文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Gu H,Takahashi H.: "Towards More Practical Average Bounds on Supervised Learning." IEEE Transactions on Neural Networks. (1996)

    • Description
      「研究成果報告書概要(欧文)」より
    • Related Report
      1995 Final Research Report Summary
  • [Publications] Honzhong Gu: "Towards More Practical Average Bounds on Supervised Learning" IEEE Trans.on Neural Networks. 22pages (1996)

    • Related Report
      1995 Annual Research Report
  • [Publications] 顧漢 忠: "概念学習における学習曲線の評価" 信学技報 ニューロコンピューティング. NC95-57. 63-70 (1995)

    • Related Report
      1995 Annual Research Report
  • [Publications] 顧漢 忠: "最悪学習曲線の上限-実用的評価を目指して-" 信学技報 ニューロコンピューティング. NC95-67. 15-22 (1995)

    • Related Report
      1995 Annual Research Report
  • [Publications] 顧漢 忠: "Self-Averaging and Sample Complexity" 信学技報 ニューロコンピューティング. NC・96-1. 7pages (1996)

    • Related Report
      1995 Annual Research Report
  • [Publications] TAKAHASHI,H: "Towards Practical Bounds of Storage Capacity in Multilayer Networks" Proc.1995 International Symposium or NOLTA. 2B-5. 215-218 (1995)

    • Related Report
      1995 Annual Research Report
  • [Publications] Honzhong Gu: "Exporential or Polybnomial Learning Curves ? A case Study" Proc.1995 International Symposium or NOLTA. 2B-12. 243-246 (1995)

    • Related Report
      1995 Annual Research Report

URL: 

Published: 1994-04-01   Modified: 2016-04-21  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi