Aarhus University Seal

Muhammad Rizwan Asif

Investigation on Projection Space Pairs in Neighbor Embedding Algorithms

Research output: Contribution to book/anthology/report/proceedingArticle in proceedingsResearchpeer-review

  • Zhe Zhang, School of Electronics and Information Engineering, Xi'an Jiaotong University
  • ,
  • Chun Qi, Xi'an Jiaotong University
  • ,
  • Muhammad Rizwan Asif

In order to achieve superior performance, various projection space pairs (PSPs) in neighbor embedding (NE) algorithms are introduced aiming to satisfy manifold assumption better. However, the comparison of theses PSPs has not been given much importance in previous researches, which could be a guiding factor for choosing better PSPs before executing the whole neighbor embedding process. Besides, evaluation criterions of final results like Peak Signal to Noise Ratio (PSNR) cannot represent the exact performance of non-linear PSPs due to the non-linear back projection process. To overcome these limitations, we compare different PSPs by introducing an efficient technique using cosine similarity and histogram approach. Experimental results demonstrate the effectiveness of the proposed evaluation method. Moreover, we also identify that non-linear PSPs could obtain superior performance only if the non-linear back projection process is well handled.

Original languageEnglish
Title of host publication2018 14th IEEE International Conference on Signal Processing Proceedings, ICSP 2018
EditorsYuan Baozong, Ruan Qiuqi, Zhao Yao, An Gaoyun
Number of pages4
PublisherInstitute of Electrical and Electronics Engineers Inc.
Publication year2 Feb 2019
Article number8652441
ISBN (Electronic)9781538646724
Publication statusPublished - 2 Feb 2019
Externally publishedYes
Event14th IEEE International Conference on Signal Processing, ICSP 2018 - Beijing, China
Duration: 12 Aug 201816 Aug 2018


Conference14th IEEE International Conference on Signal Processing, ICSP 2018
SponsorBeijing Jiaotong University, IEEE Beijing Section, IET Beijing Branch
SeriesInternational Conference on Signal Processing Proceedings, ICSP

Bibliographical note

Funding Information:
This work is supported by the National Natural Science Foundation of China (Grants Nos. 61572395 and 61675161).

Publisher Copyright:
© 2018 IEEE.

    Research areas

  • Manifold learning, Neighbor embedding, Projection space pairs

See relations at Aarhus University Citationformats

ID: 299982271