arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2409.17656cs.SDcs.AIeess.AS
收藏

用于声音事件检测自监督学习的基于原型的掩码音频模型

Prototype based Masked Audio Model for Self-Supervised Learning of Sound Event Detection

University of Science and Technology of China(中国科学技术大学);Singapore Institute of Technology(新加坡科技学院)

Pengfei Cai, Yan Song, Nan Jiang, Qing Gu, Ian McLoughlin

AI总结:

本文提出基于原型的掩码音频模型(PMAM)算法,通过GMM构建帧级伪标签并采用二元交叉熵损失进行自监督学习,有效利用无标签数据,在DESED任务上以62.5%的PSDS1分数超越现有最先进模型。

摘要:

声音事件检测(SED)中的一个重大挑战是如何有效利用无标签数据,因为高昂的标注成本导致有标签数据可用性有限。半监督算法依赖有标签数据来学习无标签数据,其性能受前者的质量和规模限制。本文引入了用于SED自监督表示学习的基于原型的掩码音频模型(PMAM)算法,以更好地利用无标签数据。具体而言,从基于高斯混合模型(GMM)的原型分布建模中构建语义丰富的帧级伪标签。这些伪标签监督基于Transformer的掩码音频模型的学习,其中采用二元交叉熵损失代替广泛使用的InfoNCE损失,以提供来自不同原型的独立损失贡献,这在多个标签可能适用于无监督数据帧的实际场景中很重要。最后阶段仅用少量有标签数据进行微调,即可得到性能极高的SED模型。在使用DESED任务的同等条件测试中,我们的方法取得了62.5%的PSDS1分数,超越了当前最先进的模型,证明了所提技术的优越性。

英文摘要:

A significant challenge in sound event detection (SED) is the effective utilization of unlabeled data, given the limited availability of labeled data due to high annotation costs. Semi-supervised algorithms rely on labeled data to learn from unlabeled data, and the performance is constrained by the quality and size of the former. In this paper, we introduce the Prototype based Masked Audio Model~(PMAM) algorithm for self-supervised representation learning in SED, to better exploit unlabeled data. Specifically, semantically rich frame-level pseudo labels are constructed from a Gaussian mixture model (GMM) based prototypical distribution modeling. These pseudo labels supervise the learning of a Transformer-based masked audio model, in which binary cross-entropy loss is employed instead of the widely used InfoNCE loss, to provide independent loss contributions from different prototypes, which is important in real scenarios in which multiple labels may apply to unsupervised data frames. A final stage of fine-tuning with just a small amount of labeled data yields a very high performing SED model. On like-for-like tests using the DESED task, our method achieves a PSDS1 score of 62.5\%, surpassing current state-of-the-art models and demonstrating the superiority of the proposed technique.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑