analiza aiqbench
This paper proposes BEAM, a training-free method for detecting anomalous sounds that stores pooled sub-band features in a memory bank, matches nearest neighbours per band, and averages the scores. The authors argue this reduces normal-score variance compared with a single global nearest-neighbour match, and they report results on DCASE Task 2 benchmarks. For ML readers, the useful points are the matching design choices and the reliance on benchmark results, which will need independent replication under varied conditions.
Sub-Band Spectral Matching with Localized Score Aggregation for Robust Anomalous Sound Detection
arXiv:2603.13749v2 Announce Type: replace-cross Abstract: Detecting subtle deviations in noisy acoustic environments is central to anomalous sound detection (ASD). A common training-free ASD pipeline temporally pools frame-level representations into a band-preserving feature vector and scores anomalies using a single nearest-neighbor match. However, this global matching can inflate normal-score variance through two effects. First, when normal sounds exhibit band-wise variability, a single global neighbor forces all bands to share the same reference, increasing band-level mismatch. Second, cosine-based matching is energy-coupled, allowing a few high-energy bands to dominate score computation under normal energy fluctuations and further increase variance. We propose BEAM, which stores temporally pooled sub-band vectors in a memory bank, retrieves neighbors per sub-band, and uniformly aggregates scores to reduce normal-score variability and improve discriminability. We further introduce a parameter-free adaptive fusion to better handle diverse temporal dynamics in sub-band responses. Experiments on multiple DCASE Task 2 benchmarks show strong performance without task-specific training, robustness to noise and domain shifts, and complementary gains when combined with encoder fine-tuning.
Abstract:Detecting subtle deviations in noisy acoustic environments is central to anomalous sound detection (ASD). A common training-free ASD pipeline temporally pools frame-level representations into a band-preserving feature vector and scores anomalies using a single nearest-neighbor match. However, this global matching can inflate normal-score variance through two effects. First, when normal sounds exhibit band-wise variability, a single global neighbor forces all bands to share the same reference, increasing band-level mismatch. Second, cosine-based matching is energy-coupled, allowing a few high-energy bands to dominate score computation under normal energy fluctuations and further increase variance. We propose BEAM, which stores temporally pooled sub-band vectors in a memory bank, retrieves neighbors per sub-band, and uniformly aggregates scores to reduce normal-score variability and improve discriminability. We further introduce a parameter-free adaptive fusion to better handle diverse temporal dynamics in sub-band responses. Experiments on multiple DCASE Task 2 benchmarks show strong performance without task-specific training, robustness to noise and domain shifts, and complementary gains when combined with encoder fine-tuning.
Sound (cs.SD); Artificial Intelligence (cs.AI)
arXiv:2603.13749 [cs.SD]
(or arXiv:2603.13749v2 [cs.SD] for this version)
https://doi.org/10.48550/arXiv.2603.13749 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 256 cuvinte
Subiecte conexe
Where, How, and When: Query-Guided Grounding for Video Understanding
arXiv · 17 oct. 2026
FedPA-LoRA: Product-Aligned Framework for Mitigating Aggregation and Initialization Errors in Heterogeneous Federated LoRA
arXiv · 19 aug. 2026
When Topology Betrays Privacy: Lattice-Based Reconstruction Attacks on Secure Aggregation in Decentralized Federated Learning
arXiv · 9 sept. 2026
Towards a Statistical Understanding of Mixture-of-Experts
arXiv · 4 sept. 2026