Metraipolis › archive › dc_0f6944f287ce

FORGE: Forward-Only Test-Time Adaptation for Integer-Only Vision Models on Microcontrollers — arXiv:2609.01683

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.01683
License (as archived)
CC0-1.0
Archived
2026-10-17T04:24:55.621Z
SHA-256 of the archived text
fbb5b9e3491c0f8f3f472ed89602053b1c2040c1378de5ea826bfb924bed3802
Stable URL of this page
https://metraipolis.com/documento/dc_0f6944f287ce

How to cite

FORGE: Forward-Only Test-Time Adaptation for Integer-Only Vision Models on Microcontrollers — arXiv:2609.01683 — arXiv (en), https://arxiv.org/abs/2609.01683 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0f6944f287ce (Metraipolis archive dc_0f6944f287ce, sha256 fbb5b9e3491c0f8f3f472ed89602053b1c2040c1378de5ea826bfb924bed3802)
BibTeX
@misc{dc0f6944f287ce,
  title = {FORGE: Forward-Only Test-Time Adaptation for Integer-Only Vision Models on Microcontrollers — arXiv:2609.01683},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.01683},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0f6944f287ce; sha256 fbb5b9e3491c0f8f3f472ed89602053b1c2040c1378de5ea826bfb924bed3802},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.01683. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision models deployed on microcontrollers (MCUs) are quantized to integer-only arithmetic and run in inference-only runtimes that do not carry the machinery backpropagation needs: the standard tool for adapting a model to the distribution shift (sensor noise, blur, lighting) it meets in the field. Existing forward-only test-time adaptation (TTA) methods either run only on server- or edge-GPU-class models (not true microcontroller integer execution), or require the batch-normalization (BN) layers that integer deployment fuses away. We present a forward-only TTA method that operates on deployed, BN-folded, integer-only convolutional networks. The key observation is that fusing BN into the preceding convolution, a mandatory step for integer inference, destroys the statistics that normalization-based adaptation relies on. We restore adaptation by re-normalizing each folded convolution's per-channel output to its clean training statistics, using only forward-pass estimates. The method (i) recovers most of gradient-based TENT's accuracy gain (+20.9 vs. +24.9 points) and matches forward-only BN adaptation, while being the only method that runs on a folded integer-only model; (ii) needs to adapt only 3 of 21 layers (selected without seeing the test corruptions) to recover 93% of the benefit; (iii) survives single-sample streaming with a batch-size-scaled momentum; and (iv) generalizes across three datasets (up to 200 classes) and two architectures. We validate bit-exact int8 convolution execution and deploy on an ESP32-S3, where, measured with a Nordic PPK2 power profiler, the forward-only adaptation (a lightweight fp32 recalibration around the int8 convolutions) costs only 8.3 mJ (6.8% of inference energy) and 21.9 ms on the deployed SIMD-optimized model: forward-only adaptation is cheap on a real microcontroller.

Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-03 in cs.AR).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Hardware Architecture (cs.AR); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.01683 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T03:29:02.878Z dalla pagina https://arxiv.org/catchup/cs.AR/2026-09-03?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).