Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Concisely Explaining the Doubt: Minimum-Size Abductive Explanations for Linear Models with a Reject Option

Gleilson Pedro Fernandes, Thiago Alves Rocha

Abstract

Trustworthiness in artificial intelligence depends not only on what a model decides, but also on how it handles and explains cases in which a reliable decision cannot be made. In critical domains such as healthcare and finance, a reject option allows the model to abstain when evidence is insufficient, making it essential to explain why an instance is rejected in order to support informed human intervention. In these settings, explanations must not only be interpretable, but also faithful to the underlying model and computationally efficient enough to support real-time decision making. Abductive explanations guarantee fidelity, but their exact computation is known to be NP-hard for many classes of models, limiting their practical applicability. Computing \textbf{minimum-size} abductive explanations is an even more challenging problem, as it requires reasoning not only about fidelity but also about optimality. Prior work has addressed this challenge in restricted settings, including log-linear-time algorithms for computing minimum-size abductive explanations in linear models without rejection, as well as a polynomial-time method based on linear programming for computing abductive explanations, without guarantees of minimum size, for linear models with a reject option. In this work, we bridge these lines of research by computing minimum-size abductive explanations for linear models with a reject option. For accepted instances, we adapt the log-linear algorithm to efficiently compute optimal explanations. For rejected instances, we formulate a 0-1 integer linear programming problem that characterizes minimum-size abductive explanations of rejection. Although this formulation is NP-hard in theory, our experimental results show that it is consistently more efficient in practice than the linear-programming-based approach that does not guarantee minimum-size explanations.

Concisely Explaining the Doubt: Minimum-Size Abductive Explanations for Linear Models with a Reject Option

Abstract

Trustworthiness in artificial intelligence depends not only on what a model decides, but also on how it handles and explains cases in which a reliable decision cannot be made. In critical domains such as healthcare and finance, a reject option allows the model to abstain when evidence is insufficient, making it essential to explain why an instance is rejected in order to support informed human intervention. In these settings, explanations must not only be interpretable, but also faithful to the underlying model and computationally efficient enough to support real-time decision making. Abductive explanations guarantee fidelity, but their exact computation is known to be NP-hard for many classes of models, limiting their practical applicability. Computing \textbf{minimum-size} abductive explanations is an even more challenging problem, as it requires reasoning not only about fidelity but also about optimality. Prior work has addressed this challenge in restricted settings, including log-linear-time algorithms for computing minimum-size abductive explanations in linear models without rejection, as well as a polynomial-time method based on linear programming for computing abductive explanations, without guarantees of minimum size, for linear models with a reject option. In this work, we bridge these lines of research by computing minimum-size abductive explanations for linear models with a reject option. For accepted instances, we adapt the log-linear algorithm to efficiently compute optimal explanations. For rejected instances, we formulate a 0-1 integer linear programming problem that characterizes minimum-size abductive explanations of rejection. Although this formulation is NP-hard in theory, our experimental results show that it is consistently more efficient in practice than the linear-programming-based approach that does not guarantee minimum-size explanations.
Paper Structure (15 sections, 2 theorems, 20 equations, 1 figure, 3 tables)

This paper contains 15 sections, 2 theorems, 20 equations, 1 figure, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Background
  4. Machine Learning and Binary Classification Problems.
  5. Linear Models and Reject Option Classification.
  6. Abductive Explanations.
  7. MINABRO
  8. Explaining Rejected Instances
  9. Explaining Classified Instances
  10. Experimental Evaluation
  11. Experimental Setup
  12. Computational Efficiency
  13. Explanation Size
  14. Visual Comparison on MNIST Rejected Instances
  15. Conclusions and Future Work

Key Result

theorem 1

Let $\mathbf{x}$ be an instance such that $t_- < s(\mathbf{x}) < t_+$. Let $(z_1^\star,\ldots,z_n^\star)$ be an optimal solution of the $0$--$1$ ILP formulation defined in Equation ilp_reject, and let $E^\star=\{f_j=x_j \mid z_j^\star=1\}$. Then $E^\star$ is a minimum-size abductive explanation of r

Figures (1)

  • Figure 1: Comparison of explanations for a rejected digit in the MNIST (3 vs. 8) task. Red pixels indicate features included in the explanation. Left: Original digit, which is rejected by the trained logistic regression classifier with reject option. Center: MINABRO explanation. Right: AbLinRO explanation.

Theorems & Definitions (8)

  • definition 1: Abductive Explanation ignatiev2019abduction
  • definition 2: Minimum-Size Abductive Explanation
  • definition 3: Minimum-size abductive explanation of rejection
  • theorem 1
  • proof
  • definition 4: Minimum-size abductive explanation of classification
  • proposition 1
  • proof