Metraipolis › archive › dc_e3ec7fc9da41

A Mixed-Precision CNN-LSTM Accelerator Featuring Long-Kernel Memory Reuse and Dynamic Quantization for Sleep Staging — arXiv:2611.00660

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.00660
License (as archived)
CC0-1.0
Archived
2026-11-03T11:12:58.416Z
SHA-256 of the archived text
5e0495dd78f5b0105282235c929763644db1b3db25fa1fe9f018b4048eee459b
Stable URL of this page
https://metraipolis.com/documento/dc_e3ec7fc9da41

How to cite

A Mixed-Precision CNN-LSTM Accelerator Featuring Long-Kernel Memory Reuse and Dynamic Quantization for Sleep Staging — arXiv:2611.00660 — arXiv (en), https://arxiv.org/abs/2611.00660 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e3ec7fc9da41 (Metraipolis archive dc_e3ec7fc9da41, sha256 5e0495dd78f5b0105282235c929763644db1b3db25fa1fe9f018b4048eee459b)
BibTeX
@misc{dce3ec7fc9da41,
  title = {A Mixed-Precision CNN-LSTM Accelerator Featuring Long-Kernel Memory Reuse and Dynamic Quantization for Sleep Staging — arXiv:2611.00660},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.00660},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e3ec7fc9da41; sha256 5e0495dd78f5b0105282235c929763644db1b3db25fa1fe9f018b4048eee459b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00660. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

This letter presents a mixed-precision CNN-LSTM accelerator for EEG-based sleep staging. To efficiently execute long temporal convolutions, a serpentine dataflow propagates activations between neighboring processing elements, reducing memory accesses by 6.2 times and 2.25 times compared with output- and weight-stationary dataflows, respectively. A hybrid quantization scheme applies runtime dynamic quantization to the LSTM and MLP layers and assigns mixed precision on a layer-by-layer basis, reducing model size by 4.16 times. Combined with a bit-flexible MAC architecture and channel-grouped memory scheduling, the accelerator supports INT2/4/8/16 inference. Fabricated in 65 nm CMOS, the accelerator consumes 0.40 mW and achieves a 258 times speedup over an iso-frequency 32b RISC-V processor. Evaluated across all 62 subjects in the public MASS-SS3 dataset using 31-fold leave-two-subject-out cross-validation, the mixed-precision approach with hybrid quantization maintains 5-class accuracy within 0.15% of the floating-point baseline when optimized for general sleep staging and N3 sensitivity within 0.51% when optimized for deep-sleep detection. The developed techniques are broadly applicable to EEG-based DL workloads.

Tema degli esploratori: calcolo e infrastrutture (elenco arXiv cs.AR, nuovi lavori del giorno).
Categorie arXiv: Hardware Architecture (cs.AR).
Pagina del lavoro: https://arxiv.org/abs/2611.00660 — 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 2026-11-03T11:00:22.163Z dalla pagina https://arxiv.org/list/cs.AR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).