---
_id: '55311'
abstract:
- lang: eng
  text: Addressing the limitations of individual attribution scores via the Shapley
    value (SV), the field of explainable AI (XAI) has recently explored intricate
    interactions of features or data points. In particular, extensions of the SV,
    such as the Shapley Interaction Index (SII), have been proposed as a measure to
    still benefit from the axiomatic basis of the SV. However, similar to the SV,
    their exact computation remains computationally prohibitive. Hence, we propose
    with SVARM-IQ a sampling-based approach to efficiently approximate Shapley-based
    interaction indices of any order. SVARM-IQ can be applied to a broad class of
    interaction indices, including the SII, by leveraging a novel stratified representation.
    We provide non-asymptotic theoretical guarantees on its approximation quality
    and empirically demonstrate that SVARM-IQ achieves state-of-the-art estimation
    results in practical XAI scenarios on different model classes and application
    domains.
author:
- first_name: Patrick
  full_name: Kolpaczki, Patrick
  last_name: Kolpaczki
- first_name: Maximilian
  full_name: Muschalik, Maximilian
  last_name: Muschalik
- first_name: Fabian
  full_name: Fumagalli, Fabian
  id: '93420'
  last_name: Fumagalli
- first_name: Barbara
  full_name: Hammer, Barbara
  last_name: Hammer
- first_name: Eyke
  full_name: Huellermeier, Eyke
  id: '48129'
  last_name: Huellermeier
citation:
  ama: 'Kolpaczki P, Muschalik M, Fumagalli F, Hammer B, Huellermeier E. SVARM-IQ:
    Efficient Approximation of Any-order Shapley Interactions through Stratification.
    In: <i>Proceedings of The 27th International Conference on Artificial Intelligence
    and Statistics (AISTATS)</i>. Vol 238. Proceedings of Machine Learning Research.
    PMLR; 2024:3520–3528.'
  apa: 'Kolpaczki, P., Muschalik, M., Fumagalli, F., Hammer, B., &#38; Huellermeier,
    E. (2024). SVARM-IQ: Efficient Approximation of Any-order Shapley Interactions
    through Stratification. <i>Proceedings of The 27th International Conference on
    Artificial Intelligence and Statistics (AISTATS)</i>, <i>238</i>, 3520–3528.'
  bibtex: '@inproceedings{Kolpaczki_Muschalik_Fumagalli_Hammer_Huellermeier_2024,
    series={Proceedings of Machine Learning Research}, title={SVARM-IQ: Efficient
    Approximation of Any-order Shapley Interactions through Stratification}, volume={238},
    booktitle={Proceedings of The 27th International Conference on Artificial Intelligence
    and Statistics (AISTATS)}, publisher={PMLR}, author={Kolpaczki, Patrick and Muschalik,
    Maximilian and Fumagalli, Fabian and Hammer, Barbara and Huellermeier, Eyke},
    year={2024}, pages={3520–3528}, collection={Proceedings of Machine Learning Research}
    }'
  chicago: 'Kolpaczki, Patrick, Maximilian Muschalik, Fabian Fumagalli, Barbara Hammer,
    and Eyke Huellermeier. “SVARM-IQ: Efficient Approximation of Any-Order Shapley
    Interactions through Stratification.” In <i>Proceedings of The 27th International
    Conference on Artificial Intelligence and Statistics (AISTATS)</i>, 238:3520–3528.
    Proceedings of Machine Learning Research. PMLR, 2024.'
  ieee: 'P. Kolpaczki, M. Muschalik, F. Fumagalli, B. Hammer, and E. Huellermeier,
    “SVARM-IQ: Efficient Approximation of Any-order Shapley Interactions through Stratification,”
    in <i>Proceedings of The 27th International Conference on Artificial Intelligence
    and Statistics (AISTATS)</i>, 2024, vol. 238, pp. 3520–3528.'
  mla: 'Kolpaczki, Patrick, et al. “SVARM-IQ: Efficient Approximation of Any-Order
    Shapley Interactions through Stratification.” <i>Proceedings of The 27th International
    Conference on Artificial Intelligence and Statistics (AISTATS)</i>, vol. 238,
    PMLR, 2024, pp. 3520–3528.'
  short: 'P. Kolpaczki, M. Muschalik, F. Fumagalli, B. Hammer, E. Huellermeier, in:
    Proceedings of The 27th International Conference on Artificial Intelligence and
    Statistics (AISTATS), PMLR, 2024, pp. 3520–3528.'
date_created: 2024-07-18T09:39:14Z
date_updated: 2025-09-11T16:22:30Z
department:
- _id: '660'
intvolume: '       238'
language:
- iso: eng
page: 3520–3528
project:
- _id: '109'
  name: 'TRR 318: TRR 318 - Erklärbarkeit konstruieren'
- _id: '117'
  name: 'TRR 318 - C: TRR 318 - Project Area C'
- _id: '126'
  name: 'TRR 318 - C3: TRR 318 - Subproject C3'
publication: Proceedings of The 27th International Conference on Artificial Intelligence
  and Statistics (AISTATS)
publisher: PMLR
series_title: Proceedings of Machine Learning Research
status: public
title: 'SVARM-IQ: Efficient Approximation of Any-order Shapley Interactions through
  Stratification'
type: conference
user_id: '93420'
volume: 238
year: '2024'
...