Explainable Artificial Intelligence: an Overview on Hybrid Models

EasyChair Preprint 14732

Abstract

The increasing integration of Artificial Intelligence (AI) in various critical areas highlights the need to
both achieve accuracy in predictions and understand the logic behind them for proper decision making.
Explainable Artificial Intelligence (XAI) addresses this challenge, balancing the complexity of models
with the necessary transparency and interpretability. Hybrid models, by integrating the accuracy of
black-box models with the transparency of interpretable ones, represent a promising avenue in the move
towards more understandable, accurate and reliable systems in AI, encouraging their safe, ethical and
responsible adoption in diverse real-world applications. This paper provides an exploration of hybrid
models in XAI, elaborating on key concepts and offering a classification based on interpretability. In
addition to describing the construction of these models, it reviews advances in the literature and identifies
future directions.

Keyphrases: Explainable Artificial Intelligence, XAI, black-box models, hybrid models, interpretability

@booklet{EasyChair:14732,
  author    = {Gabriel Quesada Álvarez and María José del Jesus Díaz and Pedro González García},
  title     = {Explainable Artificial Intelligence: an Overview on Hybrid Models},
  howpublished = {EasyChair Preprint 14732},
  year      = {EasyChair, 2024}}