(1) network structure:
NN_FLM(
  (_NN_FLM__fea_ext_net): ModelM7(
    (conv1): Conv2d(1, 48, kernel_size=(7, 7), stride=(1, 1), bias=False)
    (conv1_bn): BatchNorm2d(48, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (conv2): Conv2d(48, 96, kernel_size=(7, 7), stride=(1, 1), bias=False)
    (conv2_bn): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (conv3): Conv2d(96, 144, kernel_size=(7, 7), stride=(1, 1), bias=False)
    (conv3_bn): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (conv4): Conv2d(144, 192, kernel_size=(7, 7), stride=(1, 1), bias=False)
    (conv4_bn): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (fc1): Linear(in_features=3072, out_features=10, bias=False)
    (fc1_bn): BatchNorm1d(10, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  )
  (_NN_FLM__BFR_net): BinaryFuzzyRelationNetwork()
  (_NN_FLM__FP_loss): FuzzyPermissibleLoss(alpha=0.2, beta=0.8)
  (_NN_FLM__L2_regular): L2NormRegularization()
)

(2) network parameters:
para 1	torch.Size([48, 1, 7, 7])	_NN_FLM__fea_ext_net.conv1.weight
para 2	torch.Size([48])	_NN_FLM__fea_ext_net.conv1_bn.weight
para 3	torch.Size([48])	_NN_FLM__fea_ext_net.conv1_bn.bias
para 4	torch.Size([96, 48, 7, 7])	_NN_FLM__fea_ext_net.conv2.weight
para 5	torch.Size([96])	_NN_FLM__fea_ext_net.conv2_bn.weight
para 6	torch.Size([96])	_NN_FLM__fea_ext_net.conv2_bn.bias
para 7	torch.Size([144, 96, 7, 7])	_NN_FLM__fea_ext_net.conv3.weight
para 8	torch.Size([144])	_NN_FLM__fea_ext_net.conv3_bn.weight
para 9	torch.Size([144])	_NN_FLM__fea_ext_net.conv3_bn.bias
para 10	torch.Size([192, 144, 7, 7])	_NN_FLM__fea_ext_net.conv4.weight
para 11	torch.Size([192])	_NN_FLM__fea_ext_net.conv4_bn.weight
para 12	torch.Size([192])	_NN_FLM__fea_ext_net.conv4_bn.bias
para 13	torch.Size([10, 3072])	_NN_FLM__fea_ext_net.fc1.weight
para 14	torch.Size([10])	_NN_FLM__fea_ext_net.fc1_bn.weight
para 15	torch.Size([10])	_NN_FLM__fea_ext_net.fc1_bn.bias

(3) trade-off parameters:
gamma_FPL=	1.0
gamma_fea_ext_net_l2reg=	0.1

(4)
number of parameter in network=	15
number of parameter in optimizer=	15
