arXiv:2606.19410v1 Announce Type: cross Abstract: Signed pairwise interaction scores fundamentally conflate uniqueness (U), redundancy (R), and synergy (S). We prove this on a minimal 3-way XOR structural causal model: faithful indices such as Shapley-Taylor return zero per pair, whereas projective indices such as Shapley Interaction spread the third-order effect into pair scalars that conflate the three mechanisms. We introduce Stochastic Hi-Fi, a post-hoc, retraining-free predictability decomposition that estimates per-feature U/R/S profiles by interventional masked inference. The estimator
This research addresses a long-standing challenge in explainable AI regarding the precise attribution of feature importance, particularly for complex interactions.
Improving the interpretability of AI models is crucial for their adoption in high-stakes environments and for developing more robust and fair systems.
The introduction of Stochastic Hi-Fi offers a new method to decompose and understand feature interactions, potentially enhancing model debugging and trustworthiness beyond existing methods that conflate distinct influences.
- · AI researchers
- · Developers of explainable AI tools
- · Industries requiring high AI interpretability (e.g., finance, healthcare)
- · Explainable AI methods that poorly differentiate U/R/S
- · AI systems with opaque decision-making processes
More accurate and nuanced understanding of how features contribute to AI model predictions.
Improved ability to debug and refine complex AI models, leading to greater reliability and trust.
Accelerated development of AI agents that can explain their reasoning with greater fidelity, impacting diverse applications from autonomous systems to scientific discovery.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG