Confidence Estimation via Auxiliary Models

TL;DR

This paper proposes a new confidence estimation method for deep neural classifiers using the true class probability (TCP), learned via an auxiliary model, which improves failure prediction and domain adaptation tasks.

Contribution

The paper introduces TCP as a superior confidence measure and develops a learning scheme for it using auxiliary models, enhancing confidence estimation in neural networks.

Findings

TCP outperforms MCP in confidence estimation tasks.

The approach improves failure prediction accuracy.

Effective across various architectures and datasets.

Abstract

Reliably quantifying the confidence of deep neural classifiers is a challenging yet fundamental requirement for deploying such models in safety-critical applications. In this paper, we introduce a novel target criterion for model confidence, namely the true class probability (TCP). We show that TCP offers better properties for confidence estimation than standard maximum class probability (MCP). Since the true class is by essence unknown at test time, we propose to learn TCP criterion from data with an auxiliary model, introducing a specific learning scheme adapted to this context. We evaluate our approach on the task of failure prediction and of self-training with pseudo-labels for domain adaptation, which both necessitate effective confidence estimates. Extensive experiments are conducted for validating the relevance of the proposed approach in each task. We study various network architectures and experiment with small and large datasets for image classification and semantic segmentation. In every tested benchmark, our approach outperforms strong baselines.