FedLAS: Feature-Modulated Bidirectional Label Smoothing for Neural Network Calibration

2026-06-26

AI Summary

This work addresses the common miscalibration in deep neural networks, where predictive confidence often misaligns with actual accuracy, manifesting as either over- or under-confidence. To resolve this, the authors propose a plug-and-play bidirectional label smoothing method that leverages feature norms to construct a Normalized Confidence Indicator (NCI) and introduces a Bidirectional Calibration Gating (BCG) mechanism. This mechanism dynamically identifies the confidence state of each sample and adaptively adjusts the smoothing intensity accordingly. Unlike conventional label smoothing approaches—which only mitigate over-confidence and rely on fixed rules—this method enables sample-level bidirectional calibration for both over- and under-confident predictions. Extensive experiments on multiple standard and fine-grained visual benchmarks demonstrate significant reductions in Expected Calibration Error (ECE) and adaptive ECE, while preserving Top-1 accuracy.

Abstract

Deep Neural Network (DNN) classifiers suffer from poor calibration when their softmax outputs (predictive confidence) deviate from the empirical likelihoods. This manifests itself as either overconfident incorrect predictions or under-confident correct predictions. Label smoothing (LS) enhances model calibration by introducing entropy regularization during training through redistributing probability mass from the ground-truth label to the remaining classes. LS, including Margin-based LS (MbLS), have restrictive assumptions: they rely on predefined, uniform smoothing rules and only tackle overconfidence. In reality, samples exhibit diverse characteristics, such as difficulty/ambiguity, that interact with the evolving nature of the model being trained. In training, samples may have various degrees of under- or overconfidence. To overcome this, a mechanism that identifies the specific confidence state of each sample and determines the appropriate degree of smoothing in each training step is needed, tailoring the adjustment to the individual sample. We propose FedLAS: Feature-Modulated Bidirectional Label Smoothing, a plug-and-play algorithm for label smoothing-based losses. In FedLAS, we introduce a Feature Norm-based Confidence Indicator (NCI) to control smoothing and a Bidirectional Calibration Gating (BCG) module to detect both over and under-confidence. Our algorithm can be integrated with LS and MbLS based losses when applied to standard DNNs, enhancing performance. Extensive experiments on standard and fine-grained high-resolution vision benchmarks show that FedLAS consistently improves calibration compared to modern baselines, reducing Expected Calibration Error (ECE) and Adaptive ECE while maintaining Top-1 accuracy. Code: github.com/nadarasarbahavan/FEDLAS

Problem

• model calibration
• label smoothing
• overconfidence
• under-confidence
• confidence estimation

Innovation

• label smoothing
• model calibration
• bidirectional confidence adjustment
• feature-modulated smoothing
• confidence indicator