Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2512.13905 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914202963148800 |
|---|---|
| author | Sharify, Hossein Raoufi, Behnam Ramezani, Mahdy Hajsadeghi, Khosrow Shouraki, Saeed Bagheri |
| author_facet | Sharify, Hossein Raoufi, Behnam Ramezani, Mahdy Hajsadeghi, Khosrow Shouraki, Saeed Bagheri |
| contents | We present a compact, quantization-ready acoustic scene classification (ASC) framework that couples an efficient student network with a learned teacher ensemble and knowledge distillation. The student backbone uses stacked depthwise-separable "expand-depthwise-project" blocks with global response normalization to stabilize training and improve robustness to device and noise variability, while a global pooling head yields class logits for efficient edge inference. To inject richer inductive bias, we assemble a diverse set of teacher models and learn two complementary fusion heads: z1, which predicts per-teacher mixture weights using a student-style backbone, and z2, a lightweight MLP that performs per-class logit fusion. The student is distilled from the ensemble via temperature-scaled soft targets combined with hard labels, enabling it to approximate the ensemble's decision geometry with a single compact model. Evaluated on the TAU Urban Acoustic Scenes 2022 Mobile benchmark, our approach achieves state-of-the-art (SOTA) results on the TAU dataset under matched edge-deployment constraints, demonstrating strong performance and practicality for mobile ASC. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_13905 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Ensemble-Guided Distillation for Compact and Robust Acoustic Scene Classification on Edge Devices Sharify, Hossein Raoufi, Behnam Ramezani, Mahdy Hajsadeghi, Khosrow Shouraki, Saeed Bagheri Sound We present a compact, quantization-ready acoustic scene classification (ASC) framework that couples an efficient student network with a learned teacher ensemble and knowledge distillation. The student backbone uses stacked depthwise-separable "expand-depthwise-project" blocks with global response normalization to stabilize training and improve robustness to device and noise variability, while a global pooling head yields class logits for efficient edge inference. To inject richer inductive bias, we assemble a diverse set of teacher models and learn two complementary fusion heads: z1, which predicts per-teacher mixture weights using a student-style backbone, and z2, a lightweight MLP that performs per-class logit fusion. The student is distilled from the ensemble via temperature-scaled soft targets combined with hard labels, enabling it to approximate the ensemble's decision geometry with a single compact model. Evaluated on the TAU Urban Acoustic Scenes 2022 Mobile benchmark, our approach achieves state-of-the-art (SOTA) results on the TAU dataset under matched edge-deployment constraints, demonstrating strong performance and practicality for mobile ASC. |
| title | Ensemble-Guided Distillation for Compact and Robust Acoustic Scene Classification on Edge Devices |
| topic | Sound |
| url | https://arxiv.org/abs/2512.13905 |