Weakly-Supervised Residual Evidential Learning for Multi-Instance Uncertainty Estimation
Pei Liu, Luping Ji
TL;DR
This paper introduces Multi-Instance Uncertainty Estimation (MIUE) and proposes Multi-Instance Residual Evidential Learning (MIREL) as a principled, weakly-supervised baseline. It derives a theoretically motivated instance estimator from bag-level predictions using the Fundamental Theorem of Symmetric Functions and complements it with a residual, Dirichlet-based instance uncertainty model learned under weak supervision. The approach jointly models bag- and instance-level predictive distributions using Dirichlet posteriors and optimizes with a Fisher Information-based evidential loss plus a residual evidential loss and a regularization term. Across MNIST-bags, CIFAR10-bags, and CAMELYON16, MIREL consistently improves uncertainty estimation for both bag- and instance-level tasks, often outperforming strong UE baselines and enhancing existing MIL networks with a single forward pass. The work advances reliable decision-making in weakly-labeled MIL settings and provides code for replication.
Abstract
Uncertainty estimation (UE), as an effective means of quantifying predictive uncertainty, is crucial for safe and reliable decision-making, especially in high-risk scenarios. Existing UE schemes usually assume that there are completely-labeled samples to support fully-supervised learning. In practice, however, many UE tasks often have no sufficiently-labeled data to use, such as the Multiple Instance Learning (MIL) with only weak instance annotations. To bridge this gap, this paper, for the first time, addresses the weakly-supervised issue of Multi-Instance UE (MIUE) and proposes a new baseline scheme, Multi-Instance Residual Evidential Learning (MIREL). Particularly, at the fine-grained instance UE with only weak supervision, we derive a multi-instance residual operator through the Fundamental Theorem of Symmetric Functions. On this operator derivation, we further propose MIREL to jointly model the high-order predictive distribution at bag and instance levels for MIUE. Extensive experiments empirically demonstrate that our MIREL not only could often make existing MIL networks perform better in MIUE, but also could surpass representative UE methods by large margins, especially in instance-level UE tasks. Our source code is available at https://github.com/liupei101/MIREL.