analiza aiqbench
This research introduces PULSE, a semi-supervised multi-task framework designed to overcome the limitations of narrow training in automated insect acoustic monitoring. By leveraging self-supervised learning on unlabeled field audio and knowledge distillation from general bioacoustic models, it significantly improves species classification metrics compared to state-of-the-art baselines. The approach also generates ecologically meaningful embeddings that facilitate discovery through interactive visualization tools.
Decoding Insect Song: A Multitask Semisupervised Orthoptera Bioacoustic Classifier
arXiv:2606.13236v1 Announce Type: cross Abstract: Passive acoustic monitoring holds great promise for ecological inference, yet existing automated tools are typically narrowly trained and non-transferable. We address these limitations with PULSE, a semi-supervised, multi-task framework for Orthoptera bioacoustics, combining weakly-supervised species classification, self-supervised learning on unlabelled field audio, and knowledge distillation from a general-purpose bioacoustic model. Our domain-adapted specialist model outperforms a state-of-the-art general model across all metrics (macro F1: 0.21 vs. 0.07; AUC: 0.74 vs. 0.45; AP: 0.32 vs. 0.19), with active learning further raising F1 to 0.34 and AUC to 0.84. Beyond classification, the learned embeddings encode ecologically meaningful structure, exposed through an interactive visualisation tool for ecological discovery.
Abstract:Passive acoustic monitoring holds great promise for ecological inference, yet existing automated tools are typically narrowly trained and non-transferable. We address these limitations with PULSE, a semi-supervised, multi-task framework for Orthoptera bioacoustics, combining weakly-supervised species classification, self-supervised learning on unlabelled field audio, and knowledge distillation from a general-purpose bioacoustic model. Our domain-adapted specialist model outperforms a state-of-the-art general model across all metrics (macro F1: 0.21 vs. 0.07; AUC: 0.74 vs. 0.45; AP: 0.32 vs. 0.19), with active learning further raising F1 to 0.34 and AUC to 0.84. Beyond classification, the learned embeddings encode ecologically meaningful structure, exposed through an interactive visualisation tool for ecological discovery.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Sound (cs.SD); Applications (stat.AP)
arXiv:2606.13236 [cs.LG]
(or arXiv:2606.13236v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.13236 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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 · 206 cuvinte
Subiecte conexe
LeRobot v0.6.0: Imagine, Evaluate, Improve
Hugging Face · 7 iul. 2026
How Much Audio Is Left In An Embedding? An Inversion Audit Of Audio Encoders
arXiv · 10 oct. 2026
Google Research Shares 5 Visuals of Male Fruit Fly’s Neural Network
Quantum Zeitgeist · 15 sept. 2026
Processing and classifying bird songs using wavelet techniques and supervised learning
arXiv · 11 sept. 2026