In AI research, model multiplicity can help users better understand the diversity of AI predictions. Our new system “AI-Spectra” provides a visual dashboard to harness this concept effectively. Instead of relying on a single AI model, AI-Spectra uses multiple models—each seen as an expert—to produce predictions for the same task. This helps users see not only what different models agree or disagree on, but also why these differences occur. Gilles Eerlings (a FAIR PhD student ) and Sebe Vanbrabant where the main contributors for this work and combined machine learning, model multiplicity and visualisations that focus on the characteristics of an AI model, instead of explaining the behaviour.
To visualize this, AI-Spectra incorporates “Chernoff Bots,” cartoonish faces that represent model characteristics, allowing users to quickly grasp how different models vary. This visual approach helps manage the complexity of model multiplicity without overwhelming the user. Validated on the MNIST dataset for number recognition, AI-Spectra aims to make AI systems more transparent, trustworthy, and effective for decision-making by providing a structured and interactive comparison of model predictions.
The preprint of the paper is available on arXiv. The final paper will be published in an LNCS volume.
If you would like to cite this preprint, here is the BibTeX entry to the preprint - the final publication will become available soon:
@article{eerlings2024aispectravisualdashboardmodel,
title={AI-Spectra: A Visual Dashboard for Model Multiplicity to Enhance Informed and Transparent Decision-Making},
author={Gilles Eerlings and Sebe Vanbrabant and Jori Liesenborgs and Gustavo Rovelo Ruiz and Davy Vanacken and Kris Luyten},
year={2024},
url={https://arxiv.org/abs/2411.10490},
journal={arXiv cs.HC preprint arXiv:2411.10490},
year={2024}
}
