Exoself - Neural network lifecycle manager

The exoself is responsible for spawning, linking, and managing the lifecycle of a neural network phenotype from its genotype. It handles:

- Spawning all network processes (sensors, neurons, actuators, cortex) - Linking processes with their inputs and outputs - Weight tuning with simulated annealing - Network backup and restoration - Fitness evaluation coordination

Lifecycle

1. prep/3 - Initialize state and spawn network 2. loop/1 - Handle tuning cycles and evaluation 3. Terminate - Clean up network processes

Tuning Algorithm

The exoself uses memetic weight tuning with simulated annealing: 1. Perturb subset of weights 2. Evaluate network fitness 3. If better: keep changes, update best 4. If worse: restore previous weights 5. Reduce perturbation over attempts (annealing)