Search Sites

Transactions of the Institute of Systems, Control and Information Engineers Vol. 37 (2024), No. 10

ISIJ International
belloff
ONLINE ISSN: 2185-811X
PRINT ISSN: 1342-5668
Publisher: THE INSTITUTE OF SYSTEMS, CONTROL AND INFORMATION ENGINEERS (ISCIE)

Backnumber

  1. Vol. 37 (2024)

  2. Vol. 36 (2023)

  3. Vol. 35 (2022)

  4. Vol. 34 (2021)

  5. Vol. 33 (2020)

  6. Vol. 32 (2019)

  7. Vol. 31 (2018)

  8. Vol. 30 (2017)

  9. Vol. 29 (2016)

  10. Vol. 28 (2015)

  11. Vol. 27 (2014)

  12. Vol. 26 (2013)

  13. Vol. 25 (2012)

  14. Vol. 24 (2011)

  15. Vol. 23 (2010)

  16. Vol. 22 (2009)

  17. Vol. 21 (2008)

  18. Vol. 20 (2007)

  19. Vol. 19 (2006)

  20. Vol. 18 (2005)

  21. Vol. 17 (2004)

  22. Vol. 16 (2003)

  23. Vol. 15 (2002)

  24. Vol. 14 (2001)

  25. Vol. 13 (2000)

  26. Vol. 12 (1999)

  27. Vol. 11 (1998)

  28. Vol. 10 (1997)

  29. Vol. 9 (1996)

  30. Vol. 8 (1995)

  31. Vol. 7 (1994)

  32. Vol. 6 (1993)

  33. Vol. 5 (1992)

  34. Vol. 4 (1991)

  35. Vol. 3 (1990)

  36. Vol. 2 (1989)

  37. Vol. 1 (1988)

Transactions of the Institute of Systems, Control and Information Engineers Vol. 37 (2024), No. 10

Noise-Shaping Quantization Harmonized by Training Data for Compaction of Neural Networks

Naoki Tsubone, Yuki Minami, Masato Ishikawa

pp. 257-264

Abstract

This paper addresses the quantization of weight coefficients in trained neural networks to achieve more compact representations. To mitigate the degradation of the input-output relationship in neural networks due to quantization, we introduce the noise-shaping quantization method. This method quantizes coefficients and distributes the resulting quantization error over coefficients not yet quantized. In this paper, we proposed an approach to adjust the magnitude of the quantization error using training data. The effectiveness of the proposed method was then validated through an image classification problem. Finally, we related the quantization problem of neural network weight coefficients to the quantization problem of time signals in dynamic systems and discussed its distinctions from previous quantization techniques.

Bookmark

Share it with SNS

Article Title

Noise-Shaping Quantization Harmonized by Training Data for Compaction of Neural Networks

Trimming of Target Value Response Characteristics Using FRIT for ILQ Servo System

Takaaki Tokura, Osamu Kaneko

pp. 265-274

Abstract

Mass-produced control systems allow tolerances for each component. Therefore, the model parameters of the controlled object also have a certain tolerance, and the controller design is generally designed based on the nominal value considering the perturbation. As a result, the target value response characteristics also have a certain range of variation. The structure uses an ILQ controller and a precompensator, and the experimental data obtained in the inspection process etc. A design method that adjusts the controller using the FRIT method and restores it to the nominal target value response characteristic (this paper refers to this as trimming) is proposed. We propose this method and demonstrate its effectiveness using a numerical example.

Bookmark

Share it with SNS

Article Title

Trimming of Target Value Response Characteristics Using FRIT for ILQ Servo System

You can use this feature after you logged into the site.
Please click the button below.

Advanced Search

Article Title

Author

Abstract

Journal Title

Year

Please enter the publication date
with Christian era
(4 digits).

Please enter your search criteria.