Abstract:The beehive sound, a continuous signal produced by bees within the hive, has been found to correlate with different behavioral states of the colony, like being queenless and swarming. We investigated the possibility of identifying foraging spatial cues in this signal. We recorded a colony's sound while foraging from food sources located at three different distances from the hive, one at a time. The recordings were split into frames to obtain six statistics of their Mel Frequency Cepstral Coefficients. Then, we evaluated different autoencoding networks to obtain a latent space that allowed frames from different foraging distances to be easily differentiable. The high accuracy, silhouette score, and F1 score shown in the obtained latent spaces strongly support our approach for identifying foraging spatial cues in beehive sound activity.

What problem does this paper attempt to address?

The problem this paper attempts to solve is identifying foraging spatial cues in hive sound activities through machine learning methods. Specifically, the researchers recorded the sounds inside the hive when bees were foraging from food sources at different distances and tried to distinguish different foraging distances through these sound signals. They used autoencoding networks and statistical features of Mel-frequency cepstral coefficients (MFCCs) to achieve this goal. ### Background and Motivation 1. **Importance of Hive Sounds**: Hive sounds are continuous signals generated by the activities of bees inside the hive, known to be related to different behavioral states of the colony, such as queenlessness and swarming behavior. 2. **Communication of Foraging Behavior**: Worker bees communicate the location of resources such as nectar, pollen, and water to their peers through the waggle dance. The frequency and duration of the waggle dance can reflect the distance and direction of the food source. 3. **Existing Research**: Previous studies have shown that hive sounds can be used to detect key states of the colony, such as queenlessness, swarming behavior, and the presence of diseases or pests. However, these studies mainly focused on the overall state of the colony rather than specific foraging behaviors. ### Research Objectives 1. **Identifying Foraging Spatial Cues**: By analyzing hive sounds, identify the impact of food sources at different distances on colony activities. 2. **Method Innovation**: Use autoencoding networks and statistical features of MFCCs to extract and classify useful information from hive sounds. ### Experimental Design 1. **Data Collection**: Experiments were conducted at the Sericulture Research Institute of Yunnan Academy of Agricultural Sciences, recording the impact of food sources at different distances (100 meters, 300 meters, and 500 meters) on hive sounds. 2. **Feature Extraction**: Audio files were segmented into frames, and six statistical features of MFCCs were extracted from each frame. 3. **Model Training**: Autoencoding networks were used for pre-training, followed by fine-tuning in a supervised mode to generate new low-dimensional representations. 4. **Performance Evaluation**: The model's performance was evaluated using metrics such as Silhouette score, F1 score, and accuracy. ### Main Contributions 1. **First Successful Identification**: This is the first successful identification of foraging spatial cues from hive sound activities. 2. **Method Validation**: Validated the effectiveness of combining autoencoding networks with statistical features of MFCCs, providing new directions for further research. Through these studies, the authors hope to enhance our understanding of hive sounds as a means of communication, thereby improving the monitoring and management of colony health and behavior.

Identifying foraging spatial cues in beehive sound activity using machine learning methods

Applicability of VGGish embedding in bee colony monitoring: comparison with MFCC in colony sound classification

Soundscape Indices: New Features for Classifying Beehive Audio Samples

To bee or not to bee: Investigating machine learning approaches for beehive sound recognition

Identify The Beehive Sound Using Deep Learning

Non-Intrusive System for Honeybee Recognition Based on Audio Signals and Maximum Likelihood Classification by Autoencoder

Based investigate of beehive sound to detect air pollutants by machine learning

Semi-Supervised Audio Representation Learning for Modeling Beehive Strengths

Investigation on new Mel frequency cepstral coefficients features and hyper-parameters tuning technique for bee sound recognition

Bee Together: Joining Bee Audio Datasets for Hive Extrapolation in AI-Based Monitoring

Drone Flight Detection at an Entrance to a Beehive Based on Audio Signals

Comparative Study of Machine Learning Models for Bee Colony Acoustic Pattern Classification on Low Computational Resources

A matter of the beehive sound: Can honey bees alert the pollution out of their hives?

Automatic detection and decoding of honey bee waggle dances

Unraveling Honey Bee's Waggle Dances in Natural Conditions with Video-Based Deep Learning

VGGish-based Detection of Biological Sound Components and Their Spatio-Temporal Variations in a Subtropical Forest in Eastern China

Adaptive Representations of Sound for Automatic Insect Recognition

An artificial bee colony algorithm with a cumulative covariance matrix mechanism and its application in parameter optimization for hearing loss detection models

Small Object Tracking in High Density Crowd Scenes

Honey Bee Dance Modeling in Real-time using Machine Learning

Developing an AI-based Integrated System for Bee Health Evaluation