LGNov 26, 2025

Multiclass threshold-based classification and model evaluation

arXiv:2511.21794v11 citations
Originality Incremental advance
AI Analysis

This work addresses the problem of refining multiclass classification performance for machine learning practitioners, offering an incremental improvement through threshold tuning and new evaluation metrics.

The paper tackles multiclass classification by introducing a threshold-based framework that replaces the argmax rule with a geometric interpretation on the simplex, enabling a posteriori threshold tuning for performance improvement. Experiments show that this tuning yields gains across various networks and datasets, and a multiclass ROC analysis with a DFP score is derived as an alternative to standard methods.

In this paper, we introduce a threshold-based framework for multiclass classification that generalizes the standard argmax rule. This is done by replacing the probabilistic interpretation of softmax outputs with a geometric one on the multidimensional simplex, where the classification depends on a multidimensional threshold. This change of perspective enables for any trained classification network an \textit{a posteriori} optimization of the classification score by means of threshold tuning, as usually carried out in the binary setting, thus allowing for a further refinement of the prediction capability of any network. Our experiments show indeed that multidimensional threshold tuning yields performance improvements across various networks and datasets. Moreover, we derive a multiclass ROC analysis based on \emph{ROC clouds} -- the attainable (FPR,TPR) operating points induced by a single multiclass threshold -- and summarize them via a \emph{Distance From Point} (DFP) score to $(0,1)$. This yields a coherent alternative to standard One-vs-Rest (OvR) curves and aligns with the observed tuning gains.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes