用统计直方图提升声呐分类的参数高效微调效果
Histogram-based Parameter-efficient Tuning for Passive and Active Sonar Classification
- 通过直方图建模目标域特征分布，动态调节中间特征
- 在VTUAD数据集上达91.8%准确率，优于传统适配器的89.8%
- 适合资源受限场景，兼顾性能与参数效率
参数高效迁移学习（PETL）方法可在不微调整个模型的情况下，将大型神经网络适配到下游任务。然而，现有加性方法（如适配器）有时难以捕捉中间特征嵌入中的分布偏移。本文提出一种基于直方图的参数高效微调（HPT）技术，通过捕获目标域统计特性并调节嵌入表示。在三个被动声呐下游数据集（ShipsEar、DeepShip、VTUAD）上的实验表明，HPT优于传统适配器。尤其在VTUAD上达到91.8%准确率，相较89.8%有显著提升。对于主动声呐图像（Watertank、Turntable），HPT表现与其它PETL方法相当。此外，HPT生成的特征表示更接近全量微调模型。总体而言，HPT在节省参数的同时提供分布感知能力，是资源受限环境下迁移学习的有前景方向。代码已开源：https://github.com/Advanced-Vision-and-Learning-Lab/HLAST_DeepShip_ParameterEfficient。
原文摘要 · Abstract (English)
Parameter-efficient transfer learning (PETL) methods adapt large artificial neural networks to downstream tasks without fine-tuning the entire model. However, existing additive methods, such as adapters, sometimes struggle to capture distributional shifts in intermediate feature embeddings. We propose a novel histogram-based parameter-efficient tuning (HPT) technique that captures the statistics of the target domain and modulates the embeddings. Experimental results on three downstream passive sonar datasets (ShipsEar, DeepShip, Vessel Type Underwater Acoustic Data (VTUAD)) demonstrate that HPT outperforms conventional adapters. Notably, HPT achieves 91.8% vs. 89.8% accuracy on VTUAD. For active sonar imagery (Watertank, Turntable), HPT is competitive with other PETL methods. Furthermore, HPT yields feature representations closer to those of fully fine-tuned models. Overall, HPT balances parameter savings and provides a distribution-aware alternative to existing adapters and shows a promising direction for transfer learning in resource-constrained environments. The code is publicly available: https://github.com/Advanced-Vision-and-Learning-Lab/HLAST_DeepShip_ParameterEfficient.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。