PACE Framework Generates Feasibility-Aware Neuro-Symbolic Counterfactual Explanations

PACE combines neural predictive models with a symbolic reasoning layer to enforce domain-specific constraints during counterfactual generation. This separation ensures that generated explanations are not just mathematically minimal, but also practically actionable and realistic within human-understandable rules.