aiqbench analysis
This research provides a rigorous structural classification of how increasing voting budgets affects accuracy in binary test-time aggregation, revealing complex non-monotonic behaviors where initial gains or losses may not persist. By encoding these dynamics through a signed voting signature and moment analysis, the study offers precise mathematical tools to understand when collective decision-making improves or degrades latent population performance. These findings are critical for optimizing ensemble methods and reliability mechanisms in large-scale AI inference systems.
When Can Voting Help, Hurt, or Change Course? Exact Structure of Binary Test-Time Aggregation
arXiv:2605.05592v2 Announce Type: replace Abstract: Increasing a voting budget can improve or reduce population accuracy, and an initial gain or loss need not persist. We give a complete structural classification of these curves for binary unweighted majority under infinitely exchangeable correctness indicators, with an arbitrary latent success-probability law $\Pi$. The classification is encoded by the signed voting signature $ \omega_\Pi(B)=\int \mathbf{1}\{q(1-q)\in B\}(2q-1)\,\Pi(\,\mathrm{d} q). $ The complete odd-budget curve and this signature determine each other: the initial accuracy gives its total signed mass, and successive budget increments give its higher moments. Reflection-symmetric differences between latent laws are exactly the changes invisible to voting. A weighted total-variation constraint characterizes every attainable signature, and a minimum-mass allocation to the two reflected branches, completed by symmetric probability mass, gives every population producing its curve. Consequently, the full curve identifies the latent population precisely when no symmetric mass remains to redistribute.
Title:When Can Voting Help, Hurt, or Change Course? Exact Structure of Binary Test-Time Aggregation
Abstract:Increasing a voting budget can improve or reduce population accuracy, and an initial gain or loss need not persist. We give a complete structural classification of these curves for binary unweighted majority under infinitely exchangeable correctness indicators, with an arbitrary latent success-probability law $\Pi$. The classification is encoded by the signed voting signature $ \omega_\Pi(B)=\int \mathbf{1}\{q(1-q)\in B\}(2q-1)\,\Pi(\,\mathrm{d} q). $ The complete odd-budget curve and this signature determine each other: the initial accuracy gives its total signed mass, and successive budget increments give its higher moments. Reflection-symmetric differences between latent laws are exactly the changes invisible to voting. A weighted total-variation constraint characterizes every attainable signature, and a minimum-mass allocation to the two reflected branches, completed by symmetric probability mass, gives every population producing its curve. Consequently, the full curve identifies the latent population precisely when no symmetric mass remains to redistribute.
Machine Learning (cs.LG); Information Theory (cs.IT)
arXiv:2605.05592 [cs.LG]
(or arXiv:2605.05592v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2605.05592 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 250 words
