arXiv:2510.16462v3 Announce Type: replace Abstract: This work introduces MAYA, a sequential imitation learning model based on multi-armed bandits, designed to reproduce and predict individual bees' decisions in contextualized foraging tasks. The model accounts for bees' limited memory through a temporal window $\tau$, whose optimal value is around 7 trials, with a slight dependence on weather conditions. Experimental results on real, simulated, and complementary (mice) datasets show that MAYA (particularly with the Wasserstein distance) outperforms imitation baselines and classical statistical
The paper introduces a novel meta-bandit framework for imitation learning, particularly relevant as AI agents become more sophisticated in simulating and understanding complex behaviors.
This research provides a new model for understanding and predicting biological decision-making, which can inspire more efficient and adaptive AI agent designs, particularly for resource-constrained scenarios.
The development of MAYA demonstrates a new computational approach to replicate individual decision-making processes, particularly with limited memory, offering potential advancements in biological and AI modeling.
- · AI researchers
- · Robotics simulation labs
- · Biological modeling
- · Developers of limited-resource AI
- · Less efficient imitation learning models
- · Traditional statistical methods in behavioral science
The MAYA model could lead to more robust and biologically plausible AI agent behaviors in complex environments.
Improved understanding of optimal memory constraints in natural systems could inform the design of more energy-efficient AI hardware and algorithms.
This could contribute to the development of bio-inspired autonomous agents capable of complex decision-making in dynamic, unstructured settings, potentially influencing fields like logistics or environmental monitoring.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG