Skip to main content

Table 2 Class distribution for samples in each split of training and test set in multiclass version

From: On the utilization of deep and ensemble learning to detect milk adulteration

Dataset split Classes # Training samples # Test samples
90/10% Raw 2213 257
  Bicarbonate 419 41
  Formaldehyde 442 43
  Peroxide 417 48
  Starch 439 41
  Sucrose 431 55
75/25% Raw 1846 624
  Bicarbonate 338 122
  Formaldehyde 347 138
  Peroxide 364 101
  Starch 359 121
  Sucrose 380 106
50/50% Raw 1239 1231
  Bicarbonate 219 241
  Formaldehyde 223 262
  Peroxide 242 223
  Starch 253 227
  Sucrose 247 239
  1. In binary version, the five classes of adulterant substances are summed up as one “adulterated” class