FormulAI: Designing Rule-Based Datasets for Interpretable and Challenging Machine Learning Tasks

Abstract

In a period marked by the transformative impact of machine learning algorithms across different disciplines, challenges in achieving model interpretability persist. Existing evaluation datasets often lack transparency, obscuring the decision-making process of machine learning models, particularly in complex deep learning architectures. This opacity raises concerns spanning sectors like healthcare, emphasizing the pivotal part of explainability in breeding trust and clinging to nonsupervisory norms. While progress has been made through interpretable model developments, the absence of formalized, interpretable datasets hampers technique validation and comparison. Rule-based datasets, distinct from general synthetic datasets, offer an avenue to pretend real-world challenges while retaining interpretability. This paper presents FormulAI, a framework for generating comprehensive rule-grounded datasets, encompassing categorical and continuous features, calibrated noise, and imbalanced class distribution. Emphasizing scalability and reproducibility, these datasets function as a robust standard, fostering exploration in interpretability and robustness.

Received: 23 September 2023 | Revised: 4 March 2024 | Accepted: 15 March 2024 

Conflicts of Interest

The author declares that he has no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available at https://github.com/hextrato/FormulAI.

Author Contribution Statement

Hegler Tissot: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing - original draft, Writing - review & editing, Visualization, Supervision, Project administration.