Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Class-Level Feature Selection Method Using Feature Weighted Growing Self-Organising Maps

Andrew Starkey, Uduak Idio Akpan, Omaimah AL Hosni, Yaseen Pullissery

TL;DR

The paper addresses the limitation of global feature selection by introducing FWGSOM, a class-level FS method that leverages Growing Self-Organising Maps to identify features relevant to each class. FWGSOM computes class- and node-level feature relevance using a Samples Hit Matrix and similarity/dissimilarity matrices, enabling explainable feature attribution. Empirical results show FWGSOM achieving 100% feature selection accuracy on synthetic datasets with known class relevance and improving classification performance on real-world datasets, while also maintaining a low computational footprint. The method advances explainable AI in feature selection by providing per-class relevance insights and demonstrates robustness to varying interclass distances with sustainability considerations.

Abstract

There have been several attempts to develop Feature Selection (FS) algorithms capable of identifying features that are relevant in a dataset. Although in certain applications the FS algorithms can be seen to be successful, they have similar basic limitations. In all cases, the global feature selection algorithms seek to select features that are relevant and common to all classes of the dataset. This is a major limitation since there could be features that are specifically useful for a particular class while irrelevant for other classes, and full explanation of the relationship at class level therefore cannot be determined. While the inclusion of such features for all classes could cause improved predictive ability for the relevant class, the same features could be problematic for other classes. In this paper, we examine this issue and also develop a class-level feature selection method called the Feature Weighted Growing Self-Organising Map (FWGSOM). The proposed method carries out feature analysis at class level which enhances its ability to identify relevant features for each class. Results from experiments indicate that our method performs better than other methods, gives explainable results at class level, and has a low computational footprint when compared to other methods.

Class-Level Feature Selection Method Using Feature Weighted Growing Self-Organising Maps

TL;DR

The paper addresses the limitation of global feature selection by introducing FWGSOM, a class-level FS method that leverages Growing Self-Organising Maps to identify features relevant to each class. FWGSOM computes class- and node-level feature relevance using a Samples Hit Matrix and similarity/dissimilarity matrices, enabling explainable feature attribution. Empirical results show FWGSOM achieving 100% feature selection accuracy on synthetic datasets with known class relevance and improving classification performance on real-world datasets, while also maintaining a low computational footprint. The method advances explainable AI in feature selection by providing per-class relevance insights and demonstrates robustness to varying interclass distances with sustainability considerations.

Abstract

There have been several attempts to develop Feature Selection (FS) algorithms capable of identifying features that are relevant in a dataset. Although in certain applications the FS algorithms can be seen to be successful, they have similar basic limitations. In all cases, the global feature selection algorithms seek to select features that are relevant and common to all classes of the dataset. This is a major limitation since there could be features that are specifically useful for a particular class while irrelevant for other classes, and full explanation of the relationship at class level therefore cannot be determined. While the inclusion of such features for all classes could cause improved predictive ability for the relevant class, the same features could be problematic for other classes. In this paper, we examine this issue and also develop a class-level feature selection method called the Feature Weighted Growing Self-Organising Map (FWGSOM). The proposed method carries out feature analysis at class level which enhances its ability to identify relevant features for each class. Results from experiments indicate that our method performs better than other methods, gives explainable results at class level, and has a low computational footprint when compared to other methods.

Paper Structure

This paper contains 35 sections, 16 equations, 15 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Background Information
  3. Feature selection
  4. Filter methods
  5. Mutual Information
  6. F-Score method
  7. Wrapper Methods
  8. Heuristic Search Algorithms
  9. Genetic algorithm
  10. Recursive feature elimination and cross validation
  11. Embedded method
  12. Least Absolute Shrinkage and Selection Operator
  13. Tree based method
  14. Self-organizing Maps
  15. Growing Self-Organising Maps
  16. ...and 20 more sections

Figures (15)

  • Figure 1: FWGSOM Pseudocode
  • Figure 2: Visualisation for Moon$\_$D
  • Figure 3: Visualisation for Circle$\_$D
  • Figure 4: Visualisation for Blob$\_$D
  • Figure 5: Lung Data Samples Hit Matrix for 3x3 SOM
  • ...and 10 more figures