counterexample construction

Existing counterfactual explanations (CEs) in automated planning focus solely on minimal perturbations to *plans*, failing to expose high-level semantic properties of the underlying *planning problem*. Method: We propose “counterfactual scenarios” — a novel paradigm that identifies minimal modifications to the planning problem itself (e.g., action preconditions, goal conditions, or domain constraints), such that the modified problem admits a feasible plan satisfying a user-specified high-level property expressed in Linear Temporal Logic over finite traces (LTLf). Contribution/Results: We introduce the first formal framework for quantifying counterfactual scenarios over the space of planning problems, integrating planning logic with formal verification techniques. We systematically analyze computational complexity across diverse modification operations and prove that solving counterfactual scenarios incurs no higher worst-case complexity than solving the original planning problem. The framework achieves strong expressivity, interpretability, and practicality, significantly deepening the modeling capacity and broadening the applicability of counterfactual reasoning in automated planning.

This work proposes a novel paradigm that bridges the long-standing divide between testing and formal verification in traditional software validation, enabling them to synergistically enhance both efficiency and quality. Grounded in Design by Contract, the approach leverages the counterexample generation capability of SMT solvers to transform formal verification tools into an integrated engine for automated testing and repair. Within a unified framework, the method simultaneously achieves three key objectives: automatic generation of test cases for faulty programs, construction of regression test suites with full coverage for correct programs, and correctness-guaranteed program repair. This represents the first integration of verification, testing, and repair into a single cohesive methodology.

This work proposes a novel counterfactual training mechanism that integrates the objective of counterfactual generation directly into the model training process, addressing the limitations of existing methods which often produce explanations lacking data plausibility and feature manipulability—key requirements for supporting real-world decision-making. By incorporating priors on feature mutability, modeling the underlying data distribution, and employing constrained optimization, the approach guides the model to learn representations that yield both plausible and actionable counterfactual explanations. Experimental results demonstrate that the proposed method not only generates high-quality counterfactuals but also significantly enhances the model’s adversarial robustness, thereby offering dual benefits in interpretability and reliability.

This work addresses a critical gap in AI-driven mathematical reasoning by systematically investigating counterexample generation as an independent task, which has been largely overlooked in favor of theorem proving. The authors propose a symbolic mutation strategy to synthesize diverse training data and introduce an end-to-end verifiable training framework that fine-tunes large language models to produce counterexamples formally verifiable in Lean 4. Their approach integrates a multi-reward expert iteration mechanism to jointly optimize both counterexample generation and theorem-proving capabilities. Experimental results on three newly constructed benchmarks demonstrate that the proposed method significantly outperforms existing baselines, achieving substantial improvements in both the quality of generated counterexamples and training efficiency.

This work formally defines and systematically investigates data poisoning attacks against counterfactual explanations (CEs), revealing their mechanisms for significantly inflating algorithmic redress costs at instance-, subgroup-, and population-levels. We propose theoretically grounded, provably correct poisoning strategies and conduct adversarial injection experiments on state-of-the-art CE generators—including DiCE and CFProto—complemented by a multi-level cost measurement framework. Results demonstrate that current SOTA CE methods exhibit systemic vulnerabilities: poisoning induces substantial increases in counterfactual path length or complete failure of redress generation. Our contribution is twofold: (1) it provides a critical security alert regarding the robustness of explainable AI systems; and (2) it establishes the first principled analytical paradigm for data poisoning targeting counterfactual explanations—thereby advancing rigorous robustness evaluation and defense research for trustworthy AI.

This work addresses the limitations of existing counterfactual explanations in abstract argumentation, which predominantly rely on but-for tests and struggle with complex causal scenarios such as preemption and overdetermination. The paper proposes an intervention-based counterfactual reasoning framework that formalizes argument acceptability conditions as structural equations and incorporates refined counterfactual criteria from the Halpern-Pearl model of actual causality. By enabling simultaneous interventions on multiple variables and integrating witness constraints to fix key argument labels, the approach achieves precise identification of intricate causal structures within abstract argumentation frameworks. This method represents the first systematic effort to handle such complexities in this domain, demonstrating significantly enhanced expressiveness and reliability compared to current approaches.

This work addresses the tendency of existing coding agents to overlook underlying domain rules when repairing failing examples, often leading to recurring errors. The authors propose a counterexample-augmented, evolvable sketch approach: a human provides an initial sketch, which the agent implements; upon detecting a failing counterexample, the agent—after operator-confirmed rule validation—locally updates both the sketch and its prompts while preserving full provenance. Treating the sketch as a vehicle for internalizing domain rules, the method integrates regression-test gating and periodic clean reconstructions to avoid reliance on historical prompts or accumulated examples. Experiments on the CatSynth benchmark show that evolved sketches pass 19 out of 21 test cases, outperforming initial-sketch replay (15/21), while substantially reducing developer interventions (9 vs. 15) and lines of code changed (719 vs. 2,394).

This work addresses the challenge of maintaining counterfactual explanations (CFEs) under concept drift, a scenario where existing methods lack efficient update mechanisms. It introduces, for the first time, CFE maintenance in data stream settings and proposes a lightweight, model-agnostic updating strategy. By locally sampling around the original instance, the method dynamically repairs CFEs to preserve both validity and plausibility without requiring full regeneration, while ensuring proximity to the original input. Empirical evaluations on synthetic drifting data streams demonstrate that the approach consistently sustains CFE validity and local plausibility over time, achieving significantly lower computational overhead compared to naive re-generation baselines.

This work addresses the challenge of silent errors in large language models (LLMs) when translating natural language into optimization models, which often lead to subtle semantic deviations. The authors propose a reference-free, falsification-based verification framework that generates test instances through slot perturbations and leverages solvers for error detection. Drawing on duality theory, comparative statics, and polyhedral limit arguments, they design a suite of acoustic test classes—including collapse probes, forbidden limits, annihilation, and swap tests—that jointly examine directional, curvature, and asymptotic properties. This approach achieves, for the first time, zero false positives in verifying model fidelity. Evaluated on 326 real-world models, the method attains a 0.0% false positive rate—substantially outperforming threshold-based methods (54.9%)—and detects 70.0% of conditional mutants, 40.4% of which are invisible to conventional execution-based accuracy metrics.

This work addresses the limited interpretability of Computation Tree Logic (CTL) model checking results, which stems from the absence of intuitive, visualizable evidence forms—CTL counterexamples being notably harder to comprehend than Linear Temporal Logic (LTL) traces. The paper proposes a unified evidence framework for CTL over explicit-state models, capable of representing both witnesses for satisfied properties and counterexamples for violations. Minimal evidence structures are formally defined for each temporal operator, and a human-centric visualization scheme is developed by integrating formal reasoning with graphical techniques. A complete toolchain implementing this approach is presented. All theoretical claims are rigorously proven, significantly enhancing the readability and explainability of CTL model checking outcomes.