DDSC: Dynamic Dual-Signal Curriculum for Data-Efficient Acoustic Scene Classification under Domain Shift
This addresses domain shift in acoustic scene classification for limited-label scenarios, offering an incremental improvement over static curriculum methods.
The paper tackles the problem of device-induced domain shift in acoustic scene classification when labels are limited, proposing a dynamic curriculum training method that improves cross-device performance, with the largest gains on unseen-device splits under the DCASE 2024 protocol.
Acoustic scene classification (ASC) suffers from device-induced domain shift, especially when labels are limited. Prior work focuses on curriculum-based training schedules that structure data presentation by ordering or reweighting training examples from easy-to-hard to facilitate learning; however, existing curricula are static, fixing the ordering or the weights before training and ignoring that example difficulty and marginal utility evolve with the learned representation. To overcome this limitation, we propose the Dynamic Dual-Signal Curriculum (DDSC), a training schedule that adapts the curriculum online by combining two signals computed each epoch: a domain-invariance signal and a learning-progress signal. A time-varying scheduler fuses these signals into per-example weights that prioritize domain-invariant examples in early epochs and progressively emphasize device-specific cases. DDSC is lightweight, architecture-agnostic, and introduces no additional inference overhead. Under the official DCASE 2024 Task~1 protocol, DDSC consistently improves cross-device performance across diverse ASC baselines and label budgets, with the largest gains on unseen-device splits.