SAFE-IML - Sparsity-aware Feature Extraction for Interpretable Machine Learning with Two-stage Neural Network Modelling

Aug 3, 2026

Model interpretability has become a central research focus across diverse domains. This paper introduces SAFE-IML, a sparsity-aware feature extraction framework designed to enhance the interpretability of neural network models. SAFE-IML performs two stages: it first expands the feature space by generating many new features from an initial set, then applies dimensionality reduction to remove redundancy and retain the most important features. The selected features are used to train neural networks to improve interpretability of both outputs and model structure.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Introduction

Why Model Interpretability Is Important

Model Sparsity

[eprints@whiterose.ac.uk](eprints@whiterose.ac.uk)[ ](eprints@whiterose.ac.uk)[https://eprints.whiterose.ac.uk](https://eprints.whiterose.ac.uk) Universities of Leeds, Sheffield and York Deposited via The University of Sheffield. White Rose Research Online URL for this paper: [https://eprints.whiterose.ac.uk/id/eprint/229975/](https://eprints.whiterose.ac.uk/id/eprint/229975/) [Version: Accepted Version](Version: Accepted Version) Proceedings Paper: Wei, H.-L. (2025) SAFE-IML: Sparsity-aware feature extraction for interpretable machine learning with two-stage neural network modelling. In: 2025 10th International Conference on Machine Learning Technologies (ICMLT) . 2025 10th International Conference on Machine Learning Technologies (ICMLT), 23-25 May 2025, Helsinki, Finland. Institute of Electrical and Electronics Engineers (IEEE), pp. 188-194. ISBN: 9798331536732. [https://doi.org/10.1109/ICMLT65785.2025.11193419](https://doi.org/10.1109/ICMLT65785.2025.11193419) © 2025 The Author(s) . Except as otherwise noted, this author-accepted version of a paper published in 2025 10th International Conference on Machine Learning Technologies (ICMLT) is made available via the University of Sheffield Research Publications and Copyright Policy under the terms of the Creative Commons Attribution 4.0 International License (CC-BY 4.0), which permits unrestricted use, distribution and reproduction in any medium, provided the original work is properly cited. To view a copy of this licence, visit [http://creativecommons.org/licenses/by/4.0/](http://creativecommons.org/licenses/by/4.0/) Reuse This article is distributed under the terms of the Creative Commons Attribution (CC BY) licence. This licence allows you to distribute, remix, tweak, and build upon the work, even commercially, as long as you credit the authors for the original work. More information and the full terms of the licence here: [https://creativecommons.org/licenses/](https://creativecommons.org/licenses/) Takedown If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing [eprints@whiterose.ac.uk](eprints@whiterose.ac.uk) including the URL of the record and the reason for the withdrawal request. 2025 10th International Conference on Machine Learning Technologies (ICMLT 2025), Helsinki, Finland, May 23-25, 2025. Final accepted manuscript. SAFE-IML: Sparsity-Aware Feature Extraction for Interpretable Machine Learning with Two-Stage Neural Network Modelling Hua-Liang Wei 1,2 1 Department of Automatic Control and Systems Engineering School of Electrical and Electornic Engineering 2 Centre of Machine Intelligence The University of Sheffield Sheffield, S1 3JD, UK [w.hualiang@sheffield.ac.uk](w.hualiang@sheffield.ac.uk); ORCID: 0000-0002-4704-7346 Abstract—In recent years, model interpretability has attracted significantly increasing attention and research interests from different backgrounds and perspectives. This paper focuses on interpretation of machine learning models, aiming to propose anew sparsity-aware feature extraction (SAFE) approach to significantly improve the interpretability of neural network models. The SAFE method includes the following two steps: 1) the first step starts with a set of features used for training machine learning models, to generate a significantly large number of new features; 2) with the awareness that augmented feature space is usually redundant, the second step is focused on dimensionality reduction to identify the most important features. These important features will then be used to train neural network models, enabling much better interpretability of learning results, as well as models themselves. The proposed method is referred to as Sparsity-Aware Feature Extraction for Interpretable Machine Learning (SAFE-IML). Two illustrative examples are provided to demonstrate the applicability and efficacy of SAFE-IML. Keywords—machine learning, model int

FAQs

What problem does SAFE-IML address?

SAFE-IML targets improving the interpretability of machine learning models, specifically neural networks, by extracting and selecting features that reflect the most important drivers behind learning outcomes.

How does SAFE-IML work in two stages?

The first stage generates a large number of augmented features from an initial feature set. The second stage uses dimensionality reduction to handle redundancy in the augmented space and keep only the most important features.

How are the selected features used after extraction?

The important features identified through the second stage are then used to train neural network models, leading to better interpretability of learning results and of the models themselves.

Aug 3, 2026

Upload by Seraphina