[공지사항]을 빙자한 안부와 근황 
Show more

Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Soft-ECM: An extension of Evidential C-Means for complex data

Created by
  • Haebom

Author

Armel Soubeiga (LIMOS), Thomas Guyet (AISTROSIGHT), Violaine Antoine (LIMOS)

Outline

This paper points out the limitation of existing clustering algorithms based on belief functions that they cannot be applied to complex data (such as mixed data and time series data), and proposes a new algorithm, Soft-ECM, to solve this problem. Soft-ECM requires only a semi-metric to consistently position the center of uncertain clusters, and shows comparable results to existing fuzzy clustering approaches for numeric data. In addition, it shows the advantages of fuzzy clustering that combines the ability to process mixed data and a semi-metric such as DTW for time series data.

Takeaways, Limitations

•
Takeaways:
◦
Provides a clustering algorithm based on belief functions for complex data (mixed data, time series data, etc.).
◦
It shows comparable performance to existing fuzzy clustering approaches and is effective in handling mixed data and time series data.
◦
It can also be applied to non-Euclidean data using semi-metrics.
•
Limitations:
◦
The performance of the Soft-ECM algorithm may be affected by the choice of the semi-metric used. Further research may be needed to determine the optimal semi-metric choice.
◦
Additional extensive experimental validation on various types of complex data is needed.
◦
There is a lack of analysis of the computational complexity of the algorithm.
👍
Made with Slashpage