BeeVe framework uses unsupervised learning to decode honey bee buzzing

Researchers have developed BeeVe, a novel unsupervised framework designed to discover acoustic states within honey bee buzzing. This system utilizes a self-supervised Patchout Spectrogram Transformer (PaSST) and a Vector-Quantized Variational Autoencoder (VQ-VAE) to learn discrete acoustic tokens directly from unlabeled hive audio. The framework successfully differentiates between queenright and queenless bee conditions and identifies distinct sub-states within the queenless condition, demonstrating its potential for non-invasive hive health monitoring. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a novel unsupervised method for analyzing biological signals, potentially enabling new approaches to animal communication and health monitoring.

RANK_REASON Academic paper detailing a new unsupervised learning framework for bioacoustic analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• BeeVe
• honey bee
• Patchout Spectrogram Transformer (PaSST)
• Vector-Quantized Variational Autoencoder (VQ-VAE)

• paper
• other