Probabilistic Graph Neural Inference for wildfire evacuation logistics networks for low-power autonomous deployments

Описание и применение

Разработан легковесный вероятностный графовый нейронный сет для маршрутизации эвакуации в условиях пожаров на устройствах с низкой мощностью, что позволило сократить время инференса на Raspberry Pi в десять раз, сохранив при этом порядок риска маршрутов.

Задача

Standard GNN inference on a 4 GB Raspberry Pi without GPU caused latency in seconds per inference and drained the battery within an afternoon.

Результат

The probabilistic GNN achieved near‑real‑time inference on a low‑power Pi, cutting compute by an order of magnitude and enabling practical evacuation routing under uncertainty.

Рабочий процесс

• 1. Deploy Pi sensor nodes along trail network. 2. Attempt standard GNN inference → high latency, battery drain. 3. Study probabilistic graphical models and quantized inference. 4. Design a two‑layer probabilistic message‑passing GNN with Gaussian messages. 5. Implement moment‑matching aggregation to approximate distribution combination. 6. Quantize the model for low‑power execution. 7. Validate that risk‑ordering of routes is preserved and latency is acceptable.

Доказательства

When I actually tried to run a standard GNN inference stack on a 4GB Pi with no GPU, the whole thing collapsed. Latency ballooned to seconds per inference, and my battery budget evaporated within an afternoon.

Источник

The trick that made this tractable on-device was treating the aggregation as a moment‑matching operation rather than a full distributional combination. It's an approximation — you lose some higher‑order structure — but in my testing it preserved the risk‑ordering of routes almost perfectly while cutting compute by an order of magnitude.

Источник