LMU-Based Sequential Learning and Posterior Ensemble Fusion for Cross-Domain Infant Cry Classification
For healthcare monitoring, this work addresses the challenge of decoding infant cry causes across different datasets and infants, but the improvements are incremental over existing methods.
The paper tackles cross-domain infant cry classification, achieving improved macro-F1 under cross-domain evaluation with a compact LMU-based model that fuses multiple acoustic features and uses entropy-gated posterior ensemble fusion.
Decoding infant cry causes remains challenging for healthcare monitoring due to short nonstationary signals, limited annotations, and strong domain shifts across infants and datasets. We propose a compact acoustic framework that fuses mel-frequency cepstral coefficients (MFCCs), short-time Fourier transform (STFT) features, and fundamental-frequency (F0) contours within a multi-branch convolutional neural network (CNN) encoder, and models temporal dynamics using an enhanced Legendre Memory Unit (LMU). Compared to LSTMs, the LMU backbone provides stable sequence modeling with substantially fewer recurrent parameters, supporting efficient deployment. To improve cross-dataset generalization, we introduce calibrated posterior ensemble fusion with entropy-gated weighting to preserve domain-specific expertise while mitigating dataset bias. Experiments on Baby2020 and Baby Crying demonstrate improved macro-F1 under cross-domain evaluation, along with leakage aware splits and real-time feasibility for on-device monitoring.