Presentation information

General Session

General Session » GS-2 Machine learning

[4G1-GS-2j] 機械学習:要素技術

Fri. Jun 11, 2021 9:00 AM - 10:40 AM Room G (GS room 2)

座長:杉山 麿人 (国立情報学研究所)

9:00 AM - 9:20 AM

[4G1-GS-2j-01] Proposal of multi-dimensional data dimension reduction method using principle of synthetic wave

〇Komei Hiruta1, Eichi Takaya1, Satoshi Kurihara2 (1. Graduate School of Science and Technology, Keio University, 2. Faculty of Science and Technology, Keio University)

Keywords:data mining, dimension reduction, synthetic wave

Multidimensional data compression has long been active research area of interest from the perspective of reducing learning costs and summarizing data. Especially in today's big data era, it is expected to establish data compression methods with higher performance than ever before. In this study, we propose a new method of multidimensional compression that effectively utilizes the physical properties of waves. The principle of wave superposition can be applied to synthesize multiple waves. It is a physical fact that the wavelength of a superposition of waves with different wavelengths can be expressed as the harmonic mean of each wave. Applying this characteristics to the compression of multidimensional data, the composition of multiple variables can be expressed as the harmonic mean of each variable. In this way, we propose a method of dimensional compression that can compress each variable more effectively than conventional methods.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.