HarsanyiNet: Computing Accurate Shapley Values in a Single Forward Propagation

Lu Chen, Siyu Lou, Keyan Zhang, Jin Huang, Quanshi Zhang
4/4/2023
cs.LGcs.AIcs.CV

No official code found. Best implementation: csluchen/harsanyinet (17 stars, MIT, last commit 2023-05-24).

Abstract

The Shapley value is widely regarded as a trustworthy attribution metric. However, when people use Shapley values to explain the attribution of input variables of a deep neural network (DNN), it usually requires a very high computational cost to approximate relatively accurate Shapley values in real-world applications. Therefore, we propose a novel network architecture, the HarsanyiNet, which makes inferences on the input sample and simultaneously computes the exact Shapley values of the input variables in a single forward propagation. The HarsanyiNet is designed on the theoretical foundation that the Shapley value can be reformulated as the redistribution of Harsanyi interactions encoded by the network.

View on arXivView PDF

Code Implementations(1)

csluchen/harsanyinetCommunity100%
MIT

An interpretable network to compute the Shapley values in a single forward propagation.

174Python, Jupyter NotebookMay 9, 20233 years agoMIT

Similar Papers

Probabilistic Shapley Value Modeling and Inference

81% match2024

Towards Faithful Neural Network Intrinsic Interpretation with Shapley Additive Self-Attribution

80% match2023

SHAPNN: Shapley Value Regularized Tabular Neural Network

80% match2023

On the Computational Tractability of the (Many) Shapley Values

80% match2025

Accelerated Shapley Value Approximation for Data Evaluation

79% match2023

Derivative-based Shapley value for global sensitivity analysis and machine learning explainability

79% match2023

Cite this paper

@article{chen2023harsanyinet,
  title  = {HarsanyiNet: Computing Accurate Shapley Values in a Single Forward Propagation},
  author = {Lu Chen and Siyu Lou and Keyan Zhang and Jin Huang and Quanshi Zhang},
  year   = {2023},
  eprint = {2304.01811},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2304.01811v2},
  journal = {ICML 2023 2023}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2304.01811.svg)](https://codeofpaper.com/paper/2304.01811)

Discussion