Metraipolis › archive › dc_afbdf1d2f442

LAST: Looped Audio Spectrogram Transformer — arXiv:2610.01926

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.01926
License (as archived)
CC0-1.0
Archived
2026-10-07T12:38:44.062Z
SHA-256 of the archived text
c5d93c8afee832e8b664696319a5a3f0fd1d565682c5b5a417b91a230bc4d6c1
Stable URL of this page
https://metraipolis.com/documento/dc_afbdf1d2f442

How to cite

LAST: Looped Audio Spectrogram Transformer — arXiv:2610.01926 — arXiv (en), https://arxiv.org/abs/2610.01926 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_afbdf1d2f442 (Metraipolis archive dc_afbdf1d2f442, sha256 c5d93c8afee832e8b664696319a5a3f0fd1d565682c5b5a417b91a230bc4d6c1)
BibTeX
@misc{dcafbdf1d2f442,
  title = {LAST: Looped Audio Spectrogram Transformer — arXiv:2610.01926},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.01926},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_afbdf1d2f442; sha256 c5d93c8afee832e8b664696319a5a3f0fd1d565682c5b5a417b91a230bc4d6c1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.01926. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Increasing depth of transformer models improves recognition, but it comes at a substantial cost. Each additional layer requires more parameters, which makes the process computationally inefficient. We ask whether additional processing can focus on integrating features already computed. Looped Audio Spectrogram Transformer (LAST) first processes all tokens, then reuses the same blocks to refine only the class token over fixed audio features, thereby making later passes inexpensive. On AudioSet, ten-pass LAST achieves 0.345 mean average precision, exceeding a twelve-layer sequential transformer by 2.1% relative with 49.4% fewer parameters, 42% fewer multiply-accumulate operations, and 9.8% higher measured throughput. Across separately trained models, increasing the pass count from two to ten improves accuracy while adding only 1.2% computation. Further evaluations show improved robustness to temporal masking and various other auditory augmentations, with better generalization on classification tasks with music, environmental, and event sounds.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-02 in eess.AS).
Categorie arXiv: Sound (cs.SD) ; Machine Learning (cs.LG); Audio and Speech Processing (eess.AS).
Pagina del lavoro: https://arxiv.org/abs/2610.01926 — 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 2026-10-07T12:11:28.596Z dalla pagina https://arxiv.org/catchup/eess.AS/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).