Modeling Animal Behavior
with Metaheuristics

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.

Input
Observational Data
Sensors, video tracking, field records
Model
ANIMETA-MOD
Generic behavioral model from elementary actions
Simulation
ANIMETA-SMA
Multi-agent environment for model execution
Optimize
ANIMETA-ENGINE
Metaheuristic-driven parameter search
Output
Interpretable Model
Behavioral rules with quantified parameters

A New Approach to Animal Behavior Modeling

Unlike conventional deep learning approaches, ANIMETA frames behavioral modeling as an optimization problem — producing models that are both accurate and human-interpretable.

ANIMETA Suite Components

A 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.

ANIMETA-MOD
Generic behavioral model representation based on chained elementary actions with configurable parameters.
ANIMETA-SMA
Multi-agent simulation system that executes ANIMETA-MOD instances in a spatial, time-driven environment.
ANIMETA-ENGINE
Metaheuristic optimization core — selects optimal actions and parameters to best reproduce observed behavior.
ANIMETA-HIM
Human-interface module — clean, streamlined, and intuitive UI designed for non-computer-expert users such as biologists.

Modeling Workflow

From raw observational data to an interpretable behavioral model in four stages, powered by simulation-based optimization.

1

Data Collection

Gather behavioral data via sensors, video tracking, or direct field observations of the target species.

2

Model Configuration

Define the search space: select elementary actions, set parameter bounds, and choose a metaheuristic strategy.

3

Optimization Loop

ANIMETA-ENGINE runs the simulation iteratively, evaluating candidate solutions against the observed data until convergence.

4

Model Output

The best action sequence and parameters are extracted as an interpretable, reusable behavioral model.

Platform Features

Built for researchers and modelers who need interpretable, explainable animal behavior models without deep computer science expertise.

Elementary Action Library

Expandable database of primitive behavioral actions that agents can execute. Contribute new actions through the portal.

Metaheuristic Optimization

ANIMETA-ENGINE integrates multiple metaheuristic algorithms for efficient exploration of the behavioral parameter space.

Interpretable Outputs

Behavioral models are expressed as human-readable action sequences — no black-box inference, only explainable rules.

Fitness Evaluation

Configurable evaluation functions measure how closely a simulated behavior matches the observed data target trajectories.

Cross-Platform API

ANIMETA-API enables programmatic access for integration into scientific workflows, data pipelines, and external tools.

Multi-Agent Environment

Place multiple autonomous agents in a spatial environment. Each agent follows its own behavioral model with agent-to-agent interactions.

Ready to model animal behavior?

Start by exploring the platform tools or submit your own elementary action to the collaborative library.