Study on Scalable Image Representation through Dimension Reduction and Features Encoding for Memory Circulation Mechanism Modeling 


Vol. 45,  No. 7, pp. 1293-1305, Jul.  2020
10.7840/kics.2020.45.7.1293


PDF
  Abstract

This paper is a proof-of-concept study of variable image representation based on dimension reduction and feature value encoding. Methodologically, the dimensionality of the input data is reduced based on the manifold learning, and the image is represented as a variable bitstream through feature-based encoding in the latent space. In the implementation of the concept verification, the variational autoencoder is used as the manifold transformation method for dimension reduction and feature value extraction. We applied a scalar encoding that simulates the concept of quantifying sensory input data in human sensory organs. In the proof-of-concept experiment, it was confirmed that the 784-dimensional image is reduced to 10-dimensional features by manifold transform, and that the reduced 10-dimensional floating-point feature values can be converted into a bitstream of variable bits through scalar encoding. As a result, as the image is converted into a bitstream of variable bits, it is expected that it can be used as a memory consolidation function of the memory circulation mechanism.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Cite this article

[IEEE Style]

K. Kang and J. Cho, "Study on Scalable Image Representation through Dimension Reduction and Features Encoding for Memory Circulation Mechanism Modeling," The Journal of Korean Institute of Communications and Information Sciences, vol. 45, no. 7, pp. 1293-1305, 2020. DOI: 10.7840/kics.2020.45.7.1293.

[ACM Style]

Kyuchang Kang and Juphil Cho. 2020. Study on Scalable Image Representation through Dimension Reduction and Features Encoding for Memory Circulation Mechanism Modeling. The Journal of Korean Institute of Communications and Information Sciences, 45, 7, (2020), 1293-1305. DOI: 10.7840/kics.2020.45.7.1293.

[KICS Style]

Kyuchang Kang and Juphil Cho, "Study on Scalable Image Representation through Dimension Reduction and Features Encoding for Memory Circulation Mechanism Modeling," The Journal of Korean Institute of Communications and Information Sciences, vol. 45, no. 7, pp. 1293-1305, 7. 2020. (https://doi.org/10.7840/kics.2020.45.7.1293)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880