Argumentative Ensembling for Robust Recourse under Model Multiplicity

In machine learning, it is common to obtain multiple equally performing models for the same prediction task, e.g., when training neural networks with different random seeds. Model multiplicity (MM) is the situation which arises when these competing models differ in their predictions for the same input, for which ensembling is often employed to determine an aggregation of the outputs. Providing recourse recommendations via counterfactual explanations (CEs) under MM thus becomes complex, since the CE may not be valid across all models, i.e., the CEs are not robust under MM. In this work, we formalise the problem of providing recourse under MM, which we name recourse-aware ensembling (RAE). We propose the idea that under MM, CEs for each individual model should be considered alongside their predictions so that the aggregated prediction and recourse are decided in tandem. Centred around this intuition, we introduce six desirable properties for solutions to this problem. For solving RAE, we first extend existing ensembling methods, and show that they fall short in terms of property satisfaction. Then, we propose a novel argumentative ensembling method which guarantees the robustness of CEs under MM. Specifically, our method leverages computational argumentation to explicitly represent the conflicts between models and counterfactuals regarding prediction results and CE validity. It then uses argumentation semantics to resolve the conflicts and obtain the final solution, in a manner which is parametric to the chosen semantics. Email addresses: junqi.jiang@imperial.ac.uk (Junqi Jiang), a.rago@imperial.ac.uk (Antonio Rago), f.leofante@imperial.ac.uk (Francesco Leofante), f.toni@imperial.ac.uk (Francesca Toni) Preprint submitted to Elsevier June 26, 2025 allows for the specification of preferences over the models under MM, allowing further customisation of the ensemble. We then empirically demonstrate, across 3 datasets, the effectiveness of our approach in satisfying desirable properties with eight instantiations of our method with different semantics and model preferences. Keywords: Argumentation, Model Multiplicity, Counterfactual Explanations 1. Introduction The phenomenon of Model Multiplicity (MM) occurs when multiple, equally performing models give conflicting predictions for the same machine learning (ML) task [1]. These models may be obtained, e.g., from different random seeds, and may, e.g., model architectures, model types or high-level properties like fairness and robustness. This is also known as predictive multiplicity [2] or the Rashomon effect [3].