Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Self-Training the Neurochaos Learning Algorithm

Anusree M, Akhila Henry, Pramod P Nair

TL;DR

The paper tackles the challenge of learning with limited labeled data by introducing a hybrid semi-supervised framework that combines Neurochaos Learning (NL) with threshold-based Self-Training (ST). NL transforms inputs into chaos-based firing-rate features, while ST expands the labeled set using high-confidence pseudo-labels, mitigating noise and improving generalisation in low-data regimes. Evaluations on ten benchmark datasets across five classifiers show consistent macro-F1 gains, with particularly large improvements on nonlinear or imbalanced data such as Iris and Glass. The work demonstrates the effectiveness and robustness of chaos-inspired feature extraction paired with SSL, suggesting broad applicability to domains where labeling is expensive or scarce.

Abstract

In numerous practical applications, acquiring substantial quantities of labelled data is challenging and expensive, but unlabelled data is readily accessible. Conventional supervised learning methods frequently underperform in scenarios characterised by little labelled data or imbalanced datasets. This study introduces a hybrid semi-supervised learning (SSL) architecture that integrates Neurochaos Learning (NL) with a threshold-based Self-Training (ST) method to overcome this constraint. The NL architecture converts input characteristics into chaos-based ring-rate representations that encapsulate nonlinear relationships within the data, whereas ST progressively enlarges the labelled set utilising high-confidence pseudo-labelled samples. The model's performance is assessed using ten benchmark datasets and five machine learning classifiers, with 85% of the training data considered unlabelled and just 15% utilised as labelled data. The proposed Self-Training Neurochaos Learning (NL+ST) architecture consistently attains superior performance gain relative to standalone ST models, especially on limited, nonlinear and imbalanced datasets like Iris (188.66%), Wine (158.58%) and Glass Identification (110.48%). The results indicate that using chaos-based feature extraction with SSL improves generalisation, resilience, and classification accuracy in low-data contexts.

Self-Training the Neurochaos Learning Algorithm

TL;DR

The paper tackles the challenge of learning with limited labeled data by introducing a hybrid semi-supervised framework that combines Neurochaos Learning (NL) with threshold-based Self-Training (ST). NL transforms inputs into chaos-based firing-rate features, while ST expands the labeled set using high-confidence pseudo-labels, mitigating noise and improving generalisation in low-data regimes. Evaluations on ten benchmark datasets across five classifiers show consistent macro-F1 gains, with particularly large improvements on nonlinear or imbalanced data such as Iris and Glass. The work demonstrates the effectiveness and robustness of chaos-inspired feature extraction paired with SSL, suggesting broad applicability to domains where labeling is expensive or scarce.

Abstract

In numerous practical applications, acquiring substantial quantities of labelled data is challenging and expensive, but unlabelled data is readily accessible. Conventional supervised learning methods frequently underperform in scenarios characterised by little labelled data or imbalanced datasets. This study introduces a hybrid semi-supervised learning (SSL) architecture that integrates Neurochaos Learning (NL) with a threshold-based Self-Training (ST) method to overcome this constraint. The NL architecture converts input characteristics into chaos-based ring-rate representations that encapsulate nonlinear relationships within the data, whereas ST progressively enlarges the labelled set utilising high-confidence pseudo-labelled samples. The model's performance is assessed using ten benchmark datasets and five machine learning classifiers, with 85% of the training data considered unlabelled and just 15% utilised as labelled data. The proposed Self-Training Neurochaos Learning (NL+ST) architecture consistently attains superior performance gain relative to standalone ST models, especially on limited, nonlinear and imbalanced datasets like Iris (188.66%), Wine (158.58%) and Glass Identification (110.48%). The results indicate that using chaos-based feature extraction with SSL improves generalisation, resilience, and classification accuracy in low-data contexts.
Paper Structure (13 sections, 1 equation, 8 figures, 7 tables)

This paper contains 13 sections, 1 equation, 8 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Self-Training Algorithm
  4. Neurochaos Learning Algorithm
  5. Proposed Methodology
  6. Results and Discussion
  7. Random Forest
  8. AdaBoost
  9. Support Vector Machine
  10. Logistic Regression
  11. Gaussian Naive Bayes
  12. Comparative Analysis among Classifiers
  13. Conclusion

Figures (8)

  • Figure 1: NL Architecture
  • Figure 2: Flowchart illustrating the steps involved in the proposed NL+ST methodology.
  • Figure 3: Bar graph comparison of Random Forest performance across different architectures.
  • Figure 4: Bar graph comparison of AdaBoost performance across different architectures.
  • Figure 5: Bar graph comparison of Support Vector Machine performance across different architectures.
  • ...and 3 more figures