Breaking Diversity Collapse in Spiking Pseudo-Ensembles for Efficient OOD Detection in Remote Sensing

2026-08-02

AI Summary

This work addresses the challenge of achieving reliable out-of-distribution (OOD) detection with spiking neural networks (SNNs) in resource-constrained remote sensing systems. To this end, the authors propose a lightweight spiking pseudo-ensemble method that attaches multiple compact classification heads to a frozen SNN backbone. A novel “agree–disagree” objective function is introduced to explicitly encourage predictive diversity among the heads through structured input transformations, while preserving accuracy on in-distribution samples. Notably, this approach mitigates diversity collapse without requiring any external OOD data—a first in the field. Experimental results on EuroSAT, UCM, and AID benchmarks demonstrate that the method matches or even surpasses the performance of conventional deep ensembles composed of five full models, yet uses approximately 38% fewer parameters and reduces backbone inference calls by about 40%.

Abstract

Spiking Neural Networks (SNNs) are attractive for resource-constrained remote-sensing systems, but reliable out-of-distribution (OOD) detection remains challenging. Deep ensembles provide strong predictive uncertainty, yet require multiple complete models and backbone evaluations. We propose an efficient spiking pseudo-ensemble that attaches multiple lightweight classification heads to a frozen SNN backbone. Naively training these heads with cross-entropy can lead to diversity collapse, where independently parameterized heads may produce correlated predictions. To address this, we introduce an agree--disagree objective that preserves correct predictions on clean in-distribution samples while encouraging diversity on structured, uncertainty-inducing transformations of the same inputs. This provides a diversity-promoting training signal without requiring external OOD data. Experiments with Spikformer and ResNet19-SNN on EuroSAT demonstrate consistent improvements over conventionally trained pseudo-ensembles. Using three backbones with five heads each matches or improves upon a five-model deep ensemble on UCM and AID, while requiring approximately 38% fewer parameters and 40% fewer backbone evaluations. These results show that explicit diversity promotion can recover useful ensemble-style uncertainty at substantially lower deployment cost.

Problem

• Spiking Neural Networks
• Out-of-Distribution Detection
• Diversity Collapse
• Pseudo-Ensembles
• Remote Sensing

Innovation

• spiking pseudo-ensemble
• diversity collapse
• out-of-distribution detection
• agree-disagree objective
• efficient uncertainty estimation