Expand description
Federated learning for ternary agents.
Multiple populations share strategy insights without sharing raw data. Each node runs local ternary evolution, and a federated round aggregates strategy summaries across nodes using configurable aggregation methods, all while tracking a differential-privacy-style privacy budget.
Structs§
- Agent
- A single ternary agent with a strategy and fitness.
- Aggregator
- Aggregates strategy summaries from multiple nodes.
- Federated
Config - Configuration for a federated experiment.
- Federated
Experiment - Orchestrates a federated learning experiment.
- Federated
Round - One round of federated aggregation.
- Federation
Result - Structured result from a federated experiment.
- Node
- A local ternary population with its own evolution.
- Privacy
Budget - Tracks how much information has been shared (differential privacy style).
- Strategy
- A strategy is a vector of ternary values.
Enums§
- Aggregation
Method - Method for aggregating strategies across nodes.
- Ternary
- A ternary value: -1, 0, or +1.