BrainGraphNet/config.yaml at main · JamilHanouneh/BrainGraphNet · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
# BrainGraphNet Configuration File
# =================================

# Project Information
project:
  name: "BrainGraphNet"
  description: "Dynamic Graph Neural Networks for Evolving Functional Connectivity"
  version: "1.0.0"
  author: "Medical AI Research"

# Data Configuration
# data:
#   # Data source: 'synthetic', 'hcp', or 'custom'
#   source: 'hcp'
data:
  source: 'custom'
  custom_path: 'data/processed/HCP_connectivity'

  # Synthetic data settings
  synthetic:
    num_subjects: 50
    num_timepoints: 5
    num_regions: 90  # AAL atlas
    noise_level: 0.1
    disease_progression: true
    save_path: 'data/raw/synthetic'

  # HCP data settings
  hcp:
    data_path: 'data/raw/HCP'
    num_subjects: 100  # Use subset for faster training
    connectivity_type: 'correlation'  # 'correlation' or 'partial'

  # Custom data settings
  custom_path: 'data/raw/custom'

  # Brain atlas settings
  atlas:
    name: 'AAL'  # Automated Anatomical Labeling
    # num_regions: 90
    num_regions: 349  # Schaefer 400 parcels (adjust based on atlas used)
    labels_file: 'src/utils/aal_labels.csv'

  # Graph construction
  graph:
    threshold_method: 'proportional'  # 'absolute', 'proportional', 'none'
    threshold_value: 0.15  # Keep top 15% connections
    min_edge_weight: 0.1
    self_loops: false
    weighted: true

# Model Configuration
model:
  type: 'EvolveGCN'  # 'EvolveGCN', 'TemporalGCN', or 'Custom'

  # Architecture
  architecture:
    input_dim: 1  # Node features dimension (can add clinical data)
    hidden_dim: 64
    output_dim: 32
    num_layers: 3
    dropout: 0.3
    activation: 'relu'

  # EvolveGCN specific
  evolve_gcn:
    rnn_type: 'GRU'  # 'GRU' or 'LSTM'
    variant: 'H'  # 'H' (evolve hidden) or 'O' (evolve output)

  # Task configuration
  task:
    type: 'connectivity_prediction'  # 'connectivity_prediction' or 'phenotype_classification'
    prediction_horizon: 1  # Predict connectivity t+1 timesteps ahead

# Training Configuration
training:
  # Basic settings
  # num_epochs: 100
  num_epochs: 25
  batch_size: 8
  learning_rate: 0.001
  weight_decay: 0.0005

  # Data split
  train_ratio: 0.7
  val_ratio: 0.15
  test_ratio: 0.15

  # Optimization
  optimizer: 'adam'  # 'adam', 'sgd', 'adamw'
  scheduler:
    type: 'step'  # 'step', 'cosine', 'plateau'
    step_size: 30
    gamma: 0.5

  # Early stopping
  early_stopping:
    enabled: true
    patience: 15
    min_delta: 0.001

  # Checkpointing
  checkpoint:
    save_best: true
    save_freq: 10  # Save every N epochs
    max_to_keep: 3

# Evaluation Configuration
evaluation:
  metrics:
    - 'mse'
    - 'mae'
    - 'r2'
    - 'pearson_correlation'

  visualization:
    plot_predictions: true
    plot_brain_networks: true
    num_samples: 5

# System Configuration
system:
  device: 'cpu'  # 'cpu' or 'cuda'
  num_workers: 4
  seed: 42
  deterministic: true

  # Logging
  logging:
    level: 'INFO'  # 'DEBUG', 'INFO', 'WARNING', 'ERROR'
    save_dir: 'outputs/logs'
    tensorboard: true

  # Output paths
  paths:
    checkpoints: 'outputs/checkpoints'
    predictions: 'outputs/predictions'
    figures: 'outputs/figures'
    logs: 'outputs/logs'

# Reproducibility
random_seed: 42

# Citation
citation:
  paper: "BrainGraphNet: Dynamic Graph Neural Networks for Evolving Functional Connectivity"
  dataset: "Synthetic or HCP Young Adult Connectivity Matrices"
  code_repository: "https://github.com/JamilHanouneh/BrainGraphNet"