Probabilistic reasoning on microcontrollers for robust embedded AI sensing

Leslin, Jelin; Andraud, Martin; Morabito, Roberto
SENSYS 2026, ACM International Conference on Embedded Artificial Intelligence and Sensing Systems, Co-located with CPS-IoT Week 2026, 11-14 May 2026, Saint-Malo, France

Embedded AI sensing systems must detect unreliable predictions before acting, yet exact probabilistic reasoning is considered too costly for MCUs. In this work, we introduce
μPC, a hardware-aligned inference pipeline enabling exact probabilistic circuit execution via integer arithmetic and sparsity pruning. On an ESP32, μPC achieves up to 44x lower latency than log-domain computation while preserving validation accuracy, demonstrating practical on-device uncertainty checking1.

Type:
Poster / Demo
City:
Saint-Malo
Date:
2026-05-11
Department:
Communication systems
Eurecom Ref:
8729
Copyright:
© ACM, 2026. This is the author's version of the work. It is posted here by permission of ACM for your personal use. Not for redistribution. The definitive version was published in SENSYS 2026, ACM International Conference on Embedded Artificial Intelligence and Sensing Systems, Co-located with CPS-IoT Week 2026, 11-14 May 2026, Saint-Malo, France
See also:

PERMALINK : https://www.eurecom.fr/publication/8729