What question did this study set out to answer?

The research aims to develop a hybrid explainability methodology for time-series classification in cryptocurrency data, focusing on uncertainty estimation.

April 8, 2026Open Access

Explainability and uncertainty for time-series classification in cryptocurrency data: a hybrid XAI methodology based on COMTE and LEFTIST

LMLucas Rabelo de Araujo MoraisUniversidade Federal de Pernambuco TLTeresa B. LudermirUniversidade Federal de Pernambuco

Key Points

The research aims to develop a hybrid explainability methodology for time-series classification in cryptocurrency data, focusing on uncertainty estimation.
Applied machine learning models tailored for time-series classification.
Developed a novel hybrid explainability model combining COMTE and LEFTIST approaches.
Performed uncertainty analysis to evaluate predictive uncertainty of models.
The MRSQM model showed strong performance in time-series classification.
Machine learning models did not show significant differences from time-series models.
Deep learning models performed poorly, while uncertainty analyses revealed notable differences.

Abstract

Abstract The “Global Race For AI” has driven the pursuit of a strategy known as “AI for society”. One of the key outcomes of this strategy was the General Data Protection Regulation (GDPR), an European regulation enforced on May 28, 2018. This regulation significantly contributed to the rise of Explainable AI (XAI). Amidst this wave of technological innovation, AI systems often rely on black box models, making explainability crucial. However, when dealing with data involving time-series, which is often the case in financial contexts, time-series classification (TSC) algorithms can be applied. Therefore, XAI must also consider the temporal structure of data, highlighting the need for explainability techniques specifically tailored to TSC. In this context, this work applies Machine Learning (ML) models specifically designed for TSC and proposes a novel hybrid method that provides time-series-based explanations, while also addressing relevant topics in Responsible AI, such as uncertainty estimation and surrogate models. The results show that the MRSQM model achieved a strong performance, while ML tabular models did not differ significantly from time-series classification models, DL models performed poorly. Uncertainty analysis revealed notable differences in uncertainty estimation, and the COMTE-LEFTIST hybrid explainability model successfully provided hybrid explanations.

Ask AI

Helpful

Bookmark

View Full Paper