---
_id: '48777'
abstract:
- lang: eng
  text: <jats:title>Abstract</jats:title><jats:p>Explainable artificial intelligence
    has mainly focused on static learning scenarios so far. We are interested in dynamic
    scenarios where data is sampled progressively, and learning is done in an incremental
    rather than a batch mode. We seek efficient incremental algorithms for computing
    feature importance (FI). Permutation feature importance (PFI) is a well-established
    model-agnostic measure to obtain global FI based on feature marginalization of
    absent features. We propose an efficient, model-agnostic algorithm called iPFI
    to estimate this measure incrementally and under dynamic modeling conditions including
    concept drift. We prove theoretical guarantees on the approximation quality in
    terms of expectation and variance. To validate our theoretical findings and the
    efficacy of our approaches in incremental scenarios dealing with streaming data
    rather than traditional batch settings, we conduct multiple experimental studies
    on benchmark data with and without concept drift.</jats:p>
author:
- first_name: Fabian
  full_name: Fumagalli, Fabian
  last_name: Fumagalli
- first_name: Maximilian
  full_name: Muschalik, Maximilian
  last_name: Muschalik
- first_name: Eyke
  full_name: Hüllermeier, Eyke
  last_name: Hüllermeier
- first_name: Barbara
  full_name: Hammer, Barbara
  last_name: Hammer
citation:
  ama: 'Fumagalli F, Muschalik M, Hüllermeier E, Hammer B. Incremental permutation
    feature importance (iPFI): towards online explanations on data streams. <i>Machine
    Learning</i>. Published online 2023. doi:<a href="https://doi.org/10.1007/s10994-023-06385-y">10.1007/s10994-023-06385-y</a>'
  apa: 'Fumagalli, F., Muschalik, M., Hüllermeier, E., &#38; Hammer, B. (2023). Incremental
    permutation feature importance (iPFI): towards online explanations on data streams.
    <i>Machine Learning</i>. <a href="https://doi.org/10.1007/s10994-023-06385-y">https://doi.org/10.1007/s10994-023-06385-y</a>'
  bibtex: '@article{Fumagalli_Muschalik_Hüllermeier_Hammer_2023, title={Incremental
    permutation feature importance (iPFI): towards online explanations on data streams},
    DOI={<a href="https://doi.org/10.1007/s10994-023-06385-y">10.1007/s10994-023-06385-y</a>},
    journal={Machine Learning}, publisher={Springer Science and Business Media LLC},
    author={Fumagalli, Fabian and Muschalik, Maximilian and Hüllermeier, Eyke and
    Hammer, Barbara}, year={2023} }'
  chicago: 'Fumagalli, Fabian, Maximilian Muschalik, Eyke Hüllermeier, and Barbara
    Hammer. “Incremental Permutation Feature Importance (IPFI): Towards Online Explanations
    on Data Streams.” <i>Machine Learning</i>, 2023. <a href="https://doi.org/10.1007/s10994-023-06385-y">https://doi.org/10.1007/s10994-023-06385-y</a>.'
  ieee: 'F. Fumagalli, M. Muschalik, E. Hüllermeier, and B. Hammer, “Incremental permutation
    feature importance (iPFI): towards online explanations on data streams,” <i>Machine
    Learning</i>, 2023, doi: <a href="https://doi.org/10.1007/s10994-023-06385-y">10.1007/s10994-023-06385-y</a>.'
  mla: 'Fumagalli, Fabian, et al. “Incremental Permutation Feature Importance (IPFI):
    Towards Online Explanations on Data Streams.” <i>Machine Learning</i>, Springer
    Science and Business Media LLC, 2023, doi:<a href="https://doi.org/10.1007/s10994-023-06385-y">10.1007/s10994-023-06385-y</a>.'
  short: F. Fumagalli, M. Muschalik, E. Hüllermeier, B. Hammer, Machine Learning (2023).
date_created: 2023-11-10T14:15:36Z
date_updated: 2023-11-10T14:24:27Z
department:
- _id: '424'
- _id: '660'
doi: 10.1007/s10994-023-06385-y
keyword:
- Artificial Intelligence
- Software
language:
- iso: eng
publication: Machine Learning
publication_identifier:
  issn:
  - 0885-6125
  - 1573-0565
publication_status: published
publisher: Springer Science and Business Media LLC
status: public
title: 'Incremental permutation feature importance (iPFI): towards online explanations
  on data streams'
type: journal_article
user_id: '55908'
year: '2023'
...