arXiv：cs.LG· Yucheng Gong, Rui Zhou, Binbin Zeng, Qiang Ren, Hongjin Hui·· 4 小时前AI 评分37
冻结嵌入生物声学分类中的域对齐强度单调定律
A Strength-Monotonic Law for Domain Alignment in Frozen-Embedding Bioacoustic Classification
AI 导读
针对跨域蚊子物种分类，研究提出一条强度单调定律：编码器在目标任务上越强，其未见域泛化越依赖分布对齐（MMD）项，也越容易被域重平衡采样损害。在四类编码器及 HuBERT 层扫描（n=8）中，重平衡效应与编码器强度完全排序一致（Spearman -1.000），MMD 增益项在各流内单调、合并为 -0.857。
正文
Abstract:When does distribution alignment help a frozen foundation-model embedding generalize across acoustic domains? For cross-domain mosquito-species classification we report a strength-monotonic law: the stronger an encoder is on the target task, the more its unseen-domain generalization relies on a distribution-alignment (MMD) term, and the more it is harmed by domain-rebalanced sampling. Across four encoder families and a within-encoder HuBERT layer sweep (n=8), the rebalancing leg orders exactly with encoder strength (Spearman -1.000), while the MMD-benefit leg is monotonic within each stream and -0.857 pooled; fixing architecture and varying only representation strength flips the rebalancing effect from benefit to collapse. The law is actionable: a single MMD term is the sole lever on a strong encoder, so we reduce the field's default recipe to a frozen Perch 2.0 embedding, a lightweight probe, cross-entropy, one MMD, and input augmentation. The reduced recipe stays within seed noise of the full composite (BA_unseen 0.299+/-0.006 vs. 0.307+/-0.014). As boundary conditions of the same law, three community defaults (backbone fine-tuning, multi-modal fusion, and domain rebalancing) each hurt unseen-domain accuracy under a leave-domain protocol, shown with single-variable, multi-seed evidence. We present a mechanism and the recipe it explains, not a leaderboard entry.
| Subjects: | Audio and Speech Processing (eess.AS); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.09737 [eess.AS] |
| (or arXiv:2610.09737v1 [eess.AS] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09737 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Qiang Ren [view email]
[v1]
Wed, 7 Oct 2026 09:31:01 UTC (44 KB)
来源：arXiv：cs.LG · arxiv.org