Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Relax DARTS: Relaxing the Constraints of Differentiable Architecture Search for Eye Movement Recognition

Hongyu Zhu, Xin Jin, Hongchao Liao, Yan Xiang, Mounim A. El-Yacoubi, Huafeng Qin

Abstract

Eye movement biometrics is a secure and innovative identification method. Deep learning methods have shown good performance, but their network architecture relies on manual design and combined priori knowledge. To address these issues, we introduce automated network search (NAS) algorithms to the field of eye movement recognition and present Relax DARTS, which is an improvement of the Differentiable Architecture Search (DARTS) to realize more efficient network search and training. The key idea is to circumvent the issue of weight sharing by independently training the architecture parameters $α$ to achieve a more precise target architecture. Moreover, the introduction of module input weights $β$ allows cells the flexibility to select inputs, to alleviate the overfitting phenomenon and improve the model performance. Results on four public databases demonstrate that the Relax DARTS achieves state-of-the-art recognition performance. Notably, Relax DARTS exhibits adaptability to other multi-feature temporal classification tasks.

Relax DARTS: Relaxing the Constraints of Differentiable Architecture Search for Eye Movement Recognition

Abstract

Eye movement biometrics is a secure and innovative identification method. Deep learning methods have shown good performance, but their network architecture relies on manual design and combined priori knowledge. To address these issues, we introduce automated network search (NAS) algorithms to the field of eye movement recognition and present Relax DARTS, which is an improvement of the Differentiable Architecture Search (DARTS) to realize more efficient network search and training. The key idea is to circumvent the issue of weight sharing by independently training the architecture parameters to achieve a more precise target architecture. Moreover, the introduction of module input weights allows cells the flexibility to select inputs, to alleviate the overfitting phenomenon and improve the model performance. Results on four public databases demonstrate that the Relax DARTS achieves state-of-the-art recognition performance. Notably, Relax DARTS exhibits adaptability to other multi-feature temporal classification tasks.
Paper Structure (13 sections, 8 equations, 4 figures, 5 tables)

This paper contains 13 sections, 8 equations, 4 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Traditional eye movement recognition method
  4. Deep Learning based eye movement recognition method
  5. Differentiable architecture search
  6. Approach
  7. Review Relax DARTS
  8. Local Search Strategy
  9. Global Search Strategy
  10. Experiments
  11. Identification Performance of Relax DARTS
  12. Ablation Experiment
  13. Conclusion

Figures (4)

  • Figure 1: The overall flow chart of eye movement (EM) recognition using the NAS algorithm. Compared to the existing DARTS-based algorithmsDARTS2018FairDarts2019Chu2020DARTSR-, we directly utilize the searched architectures for training and evaluation without stacking of units.
  • Figure 2: The framework of the proposed Relax DARTS algorithm
  • Figure : Results of comparative experiments on the RAN database
  • Figure : Results of comparative experiments on the TEX database