Transitional Uncertainty with Layered Intermediate Predictions

TL;DR

This paper introduces TULIP, a method that improves uncertainty estimation in neural networks by preserving intermediate features, addressing limitations of current single-pass estimators especially in complex or imbalanced data scenarios.

Contribution

The paper proposes Transitional Uncertainty with Layered Intermediate Predictions (TULIP), a novel approach that preserves intermediate features to enhance uncertainty estimation in neural networks.

Findings

TULIP matches or outperforms existing methods on standard benchmarks.

TULIP is more reliable in imbalanced and complex data settings.

Preserving intermediate features improves uncertainty estimates.

Abstract

In this paper, we discuss feature engineering for single-pass uncertainty estimation. For accurate uncertainty estimates, neural networks must extract differences in the feature space that quantify uncertainty. This could be achieved by current single-pass approaches that maintain feature distances between data points as they traverse the network. While initial results are promising, maintaining feature distances within the network representations frequently inhibits information compression and opposes the learning objective. We study this effect theoretically and empirically to arrive at a simple conclusion: preserving feature distances in the output is beneficial when the preserved features contribute to learning the label distribution and act in opposition otherwise. We then propose Transitional Uncertainty with Layered Intermediate Predictions (TULIP) as a simple approach to address the shortcomings of current single-pass estimators. Specifically, we implement feature preservation by extracting features from intermediate representations before information is collapsed by subsequent layers. We refer to the underlying preservation mechanism as transitional feature preservation. We show that TULIP matches or outperforms current single-pass methods on standard benchmarks and in practical settings where these methods are less reliable (imbalances, complex architectures, medical modalities).