Skip to content

Latest commit

 

History

History
22 lines (22 loc) · 676 Bytes

bf_multi-label_classification_out_20170819-010852.org

File metadata and controls

22 lines (22 loc) · 676 Bytes

System parameters

  • Numpy random number seed: 0
  • Ratio of training data to overall data: 0.90
  • Number of epochs: 20
  • Batch size: 16
  • SAE hidden layers: 64-4-64
  • SAE activation: relu
  • SAE bias: False
  • SAE optimizer: adam
  • SAE loss: mse
  • Classifier hidden layers: 32
  • Classifier hidden layer activation: relu
  • Classifier bias: False
  • Classifier optimizer: adam
  • Classifier loss: binary_crossentropy
  • Classifier dropout rate: 0.10
  • Classifier class weight for buildings: 50.00
  • Classifier class weight for floors: 1.00

Performance

  • Accuracy (building) = 1.000000e+00
  • Accuracy (floor) = 9.000900e-01