import numpy as np
import logging
from timeit import default_timer as timer
from scipy.optimize import fmin_l_bfgs_b, basinhopping
import torch
import torch.nn.functional as F
from v1_metrics import compute_eer
import data_reader.adv_kaldi_io as ako
"""
utterance-based validation without stochastic search for threshold
important: EER does not need a threshold.
"""
## Get the same logger from main"
logger = logging.getLogger("anti-spoofing")
def validation(args, model, device, val_loader, val_scp, val_utt2label):
logger.info("Starting Validation")
val_loss, val_scores = compute_loss(model, device, val_loader)
val_preds, val_labels = utt_scores(val_scores, val_scp, val_utt2label)
val_eer = compute_eer(val_labels, val_preds)
logger.info('===> Validation set: Average loss: {:.4f}\tEER: {:.4f}\n'.format(
val_loss, val_eer))
return val_loss, val_eer
def utt_scores(scores, scp, utt2label):
"""return predictions and labels per utterance
"""
utt2label = ako.read_key_label(utt2label)
key_list = ako.read_all_key(scp)
preds, labels = [], []
idx = 0
for key in key_list:
preds.append(scores[idx])
idx += 1
labels.append(utt2label[key])
return np.array(preds), np.array(labels)
def compute_loss(model, device, data_loader):
model.eval()
loss = 0
scores = []
with torch.no_grad():
for X1, X2, target in data_loader:
X1, X2, target = X1.to(device), X2.to(device), target.to(device)
target = target.view(-1,1).float()
y = model(X1, X2)
loss += F.binary_cross_entropy(y, target, size_average=False)
scores.append(y.data.cpu().numpy())
loss /= len(data_loader.dataset) # average loss
scores = np.vstack(scores) # scores per utterance
return loss, scores
