File size: 1,247 Bytes
9efc160
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
{
  "model_name": "EEG Data Synthesis with WGAN-GP",
  "architecture": "conditional_wgan_gp",
  "latent_dim": 128,
  "num_subjects": 109,
  "num_channels": 64,
  "segment_length": 480,
  "generator_fc_layers": [256, 2048, 7680],
  "deconv_layers": [
    {"type": "ConvTranspose1d", "kernel_size": 4, "stride": 4, "dilation": 1},
    {"type": "Conv1d", "kernel_size": 3, "padding": 2, "dilation": 2},
    {"type": "Conv1d", "kernel_size": 3, "padding": 4, "dilation": 4}
  ],
  "activation": "tanh",
  "optimizer": {
    "type": "Adam",
    "beta1": 0.0,
    "beta2": 0.9,
    "lr_generator": 1e-4,
    "lr_discriminator": 5e-5
  },
  "training": {
    "epochs": 300,
    "batch_size": 42,
    "gradient_penalty_lambda": 5,
    "drift_regularization": 0.001,
    "n_critic": 1,
    "mixed_precision": true
  },
  "dataset": {
    "name": "PhysioNet EEG Motor Movement/Imagery",
    "num_subjects": 109,
    "sampling_rate": 160,
    "channels": 64,
    "segment_length": 480,
    "normalization": "[-1, 1]",
    "tasks": ["left_fist", "right_fist", "both_fists", "both_feet", "eyes_open", "eyes_closed"]
  },
  "description": "Trained conditional EEG generator (WGAN-GP) using 109 subjects from the PhysioNet EEG Motor Movement/Imagery dataset."
}