Skip to content

federation-plans.md

Latest commit

 

History

History
148 lines (116 loc) · 3 KB

federation-plans.md

File metadata and controls

148 lines (116 loc) · 3 KB

Federation Plans

Federation plans are YAML configuration files that define how federated learning sessions are conducted. They specify the participants, algorithms, models, and other parameters.

Plan Structure

A federation plan consists of several key sections:

Basic Structure

federation:
  name: "string"           # Unique federation name
  description: "string"    # Human-readable description
  
  aggregator:              # Aggregator configuration
    host: "string"
    port: integer
    
  collaborators:           # List of collaborators
    - name: "string"
      host: "string"
      port: integer
      
  algorithm:               # Federated learning algorithm
    type: "string"
    parameters: {}
    
  model:                   # Model configuration
    type: "string"
    parameters: {}
    
  monitoring:              # Monitoring configuration (optional)
    enabled: boolean
    config: {}

Algorithm Types

FedAvg (Federated Averaging)

algorithm:
  type: "fedavg"
  rounds: 10
  epochs_per_round: 5
  learning_rate: 0.01

FedOpt (Federated Optimization)

algorithm:
  type: "fedopt"
  rounds: 10
  epochs_per_round: 5
  learning_rate: 0.01
  beta1: 0.9
  beta2: 0.999

FedProx (Federated Proximal)

algorithm:
  type: "fedprox"
  rounds: 10
  epochs_per_round: 5
  learning_rate: 0.01
  mu: 0.001

Model Types

Simple Neural Network

model:
  type: "simple_nn"
  input_size: 784
  hidden_size: 128
  output_size: 10
  activation: "relu"

Convolutional Neural Network

model:
  type: "cnn"
  input_channels: 1
  input_height: 28
  input_width: 28
  num_classes: 10
  layers: [32, 64, 128]

Monitoring Configuration

monitoring:
  enabled: true
  config:
    api_port: 8080
    webui_port: 3000
    metrics_retention: "24h"
    collection_interval: "5s"

Security Configuration

mTLS (Mutual TLS)

security:
  mtls:
    enabled: true
    ca_cert: "path/to/ca.crt"
    cert: "path/to/cert.crt"
    key: "path/to/key.key"

Example Plans

See the examples directory for complete working examples:

Validation

Validate your plan before running:

fx plan validate my_federation.yaml

Best Practices

  1. Use descriptive names for federations and collaborators
  2. Set appropriate timeouts for network operations
  3. Enable monitoring for production deployments
  4. Use security features for sensitive data
  5. Test with small datasets before scaling up

Troubleshooting

Common issues and solutions:

  • Connection refused: Check host/port configuration
  • Invalid plan: Use fx plan validate to check syntax
  • Performance issues: Adjust algorithm parameters
  • Security errors: Verify certificate paths and permissions
⚡