Modeling Bumblebee Memory
This work seeks to improve our understanding of how anthropogenic disturbances affect plant–pollinator systems. Studies show that pesticides and pathogens can impair behavioral processes needed for pollinators to adaptively exploit floral resources and effectively transfer pollen among plants. However, the potential for these sublethal stressor effects on pollinator–plant interactions at the individual level to scale up into changes to the dynamics of wild plant and pollinator populations at the system level remains unclear. I developed and tested an empirically parameterized agent-based model of a bumblebee pollination system to test for effects of stressor-induced decreases in the memory capacity and information processing speed of individual foragers on bumblebee abundance, plant diversity, and pollinator–plant system stability.
You can find the published paper and model here: https://doi.org/10.1111/cobi.13754
Classifying Bumblebee Behaviors
In the past five years, researchers have made huge strides in using deep learning methods to improve the classification of human behavior. However, relatively few studies have applied these methods to non-anthropomorphic models. The goal of my research is to study how well multi-stream neural networks can be applied to temporal behavior classification in bumblebees and improve our understanding of how well these deep learning methods translate to non-anthropomorphic organisms.
The original RGB video provides a wealth of information for human eyes, but can prove immensely complex for deep learning methods.
A single video can display any number of behaviors, from searching for a flower to gathering pollen. The goal of my research is to combine deep learning temporal action detection methods with more traditional machine learning methods to fully utilize the temporal dependencies of bumblebee behavior.