ANIMETA is a research-driven platform that treats animal behavior modeling as an optimization problem. Design agents, build environments, and generate interpretable behavioral models using metaheuristics.
Unlike conventional deep learning approaches, ANIMETA frames behavioral modeling as an optimization problem — producing models that are both accurate and human-interpretable.
Propose and submit elementary actions to expand the platform's behavioral building block library. Community contributions enrich the simulation capabilities.
Open Contribution PortalInput observational data, configure optimization parameters, and submit your modeling project. ANIMETA-ENGINE uses metaheuristics to generate an interpretable behavioral model.
Open Modeling StudioDesign agents with specific behavioral models, define their environment, set spatial parameters, and launch multi-agent simulations powered by ANIMETA-SMA.
Open Simulation BuilderA complete toolchain from data input to interpretable behavioral models, built on a multi-agent simulation backbone.
The ANIMETA system was designed and validated through rigorous verification and validation procedures, ensuring reliability for scientific modeling tasks in ethology and behavioral ecology.
At its core, ANIMETA treats the behavioral modeling problem as a simulation-based optimization task: a database of elementary actions serves as the search space, and metaheuristics drive the exploration for the best action sequences and parameters.
The platform has been validated against established benchmarks and tested on real observational data, including direct field observations of Sciaena umbra (brown grouper) and controlled pig behavior studies.
From raw observational data to an interpretable behavioral model in four stages, powered by simulation-based optimization.
Gather behavioral data via sensors, video tracking, or direct field observations of the target species.
Define the search space: select elementary actions, set parameter bounds, and choose a metaheuristic strategy.
ANIMETA-ENGINE runs the simulation iteratively, evaluating candidate solutions against the observed data until convergence.
The best action sequence and parameters are extracted as an interpretable, reusable behavioral model.
Built for researchers and modelers who need interpretable, explainable animal behavior models without deep computer science expertise.
Expandable database of primitive behavioral actions that agents can execute. Contribute new actions through the portal.
ANIMETA-ENGINE integrates multiple metaheuristic algorithms for efficient exploration of the behavioral parameter space.
Behavioral models are expressed as human-readable action sequences — no black-box inference, only explainable rules.
Configurable evaluation functions measure how closely a simulated behavior matches the observed data target trajectories.
ANIMETA-API enables programmatic access for integration into scientific workflows, data pipelines, and external tools.
Place multiple autonomous agents in a spatial environment. Each agent follows its own behavioral model with agent-to-agent interactions.
Start by exploring the platform tools or submit your own elementary action to the collaborative library.