Modality-Aware Zero-Shot Pruning and Sparse Attention for Efficient Multimodal Edge Inference
This addresses the challenge of deploying accurate multimodal AI on resource-constrained edge devices, offering a practical solution for applications like sensing pipelines, though it is incremental in improving existing pruning methods.
The paper tackled the problem of inefficient multimodal edge inference under fluctuating power and sensor dropout by introducing SentryFuse, which achieved a 12.7% average accuracy improvement over baselines and reduced memory by 28.2% and latency by up to 1.63x without fine-tuning.
Edge devices increasingly run multimodal sensing pipelines that must remain accurate despite fluctuating power budgets and unpredictable sensor dropout. Existing pruning methods fail under these conditions: they generally require fine-tuning after compression, consuming over $10\times$ the deployment energy, and they assign static importance scores that are blind to which sensors are present. We present the SentryFuse framework, which addresses both challenges jointly through two key components. First, SentryGate learns modality-conditioned importance scores during training via first-order saliency supervision and then prunes attention heads and feed-forward channels at deployment without fine-tuning. Second, SentryAttend replaces dense self-attention, a key bottleneck in contemporary multimodal architectures, with sparse grouped-query attention, yielding a net 15% reduction in GFLOPs across three different multimodal architectures. Across three applications and multimodal backbones, SentryGate achieves a 12.7% average accuracy improvement over the strongest pruning baseline, and upto to 18% under modality dropout conditions. Together, SentryFuse reduces memory by 28.2% and lowers latency by up to $1.63\times$ without further fine-tuning, establishing modality-aware zero-shot compression as a practical path to multimodal intelligence on heterogeneous edge hardware.