'how correctly set loss_weights or class_weights for binary cross_entropy?
how correctly set loss_weights for method compile() or class_weights for fit(). I use binary cross_entropy. I trying train model for binary classification. My labels has two chanels [0,1][0,1] for each class. My model.summary():
Model: "functional_1"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) [(None, 256, 256, 1) 0
__________________________________________________________________________________________________
conv2d (Conv2D) (None, 256, 256, 64) 640 input_1[0][0]
__________________________________________________________________________________________________
batch_normalization (BatchNorma (None, 256, 256, 64) 256 conv2d[0][0]
__________________________________________________________________________________________________
conv2d_1 (Conv2D) (None, 256, 256, 64) 36928 batch_normalization[0][0]
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 256, 256, 64) 256 conv2d_1[0][0]
__________________________________________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 128, 128, 64) 0 batch_normalization_1[0][0]
__________________________________________________________________________________________________
conv2d_2 (Conv2D) (None, 128, 128, 128 73856 max_pooling2d[0][0]
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 128, 128, 128 512 conv2d_2[0][0]
__________________________________________________________________________________________________
conv2d_3 (Conv2D) (None, 128, 128, 128 147584 batch_normalization_2[0][0]
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 128, 128, 128 512 conv2d_3[0][0]
__________________________________________________________________________________________________
max_pooling2d_1 (MaxPooling2D) (None, 64, 64, 128) 0 batch_normalization_3[0][0]
__________________________________________________________________________________________________
conv2d_4 (Conv2D) (None, 64, 64, 256) 295168 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 64, 64, 256) 1024 conv2d_4[0][0]
__________________________________________________________________________________________________
conv2d_5 (Conv2D) (None, 64, 64, 256) 590080 batch_normalization_4[0][0]
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 64, 64, 256) 1024 conv2d_5[0][0]
__________________________________________________________________________________________________
max_pooling2d_2 (MaxPooling2D) (None, 32, 32, 256) 0 batch_normalization_5[0][0]
__________________________________________________________________________________________________
conv2d_6 (Conv2D) (None, 32, 32, 512) 1180160 max_pooling2d_2[0][0]
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 32, 32, 512) 2048 conv2d_6[0][0]
__________________________________________________________________________________________________
conv2d_7 (Conv2D) (None, 32, 32, 512) 2359808 batch_normalization_6[0][0]
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 32, 32, 512) 2048 conv2d_7[0][0]
__________________________________________________________________________________________________
max_pooling2d_3 (MaxPooling2D) (None, 16, 16, 512) 0 batch_normalization_7[0][0]
__________________________________________________________________________________________________
conv2d_8 (Conv2D) (None, 16, 16, 1024) 4719616 max_pooling2d_3[0][0]
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 16, 16, 1024) 4096 conv2d_8[0][0]
__________________________________________________________________________________________________
conv2d_9 (Conv2D) (None, 16, 16, 1024) 9438208 batch_normalization_8[0][0]
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 16, 16, 1024) 4096 conv2d_9[0][0]
__________________________________________________________________________________________________
up_sampling2d (UpSampling2D) (None, 32, 32, 1024) 0 batch_normalization_9[0][0]
__________________________________________________________________________________________________
conv2d_10 (Conv2D) (None, 32, 32, 512) 2097664 up_sampling2d[0][0]
__________________________________________________________________________________________________
concatenate (Concatenate) (None, 32, 32, 1024) 0 batch_normalization_7[0][0]
conv2d_10[0][0]
__________________________________________________________________________________________________
conv2d_11 (Conv2D) (None, 32, 32, 512) 4719104 concatenate[0][0]
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 32, 32, 512) 2048 conv2d_11[0][0]
__________________________________________________________________________________________________
conv2d_12 (Conv2D) (None, 32, 32, 512) 2359808 batch_normalization_10[0][0]
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 32, 32, 512) 2048 conv2d_12[0][0]
__________________________________________________________________________________________________
up_sampling2d_1 (UpSampling2D) (None, 64, 64, 512) 0 batch_normalization_11[0][0]
__________________________________________________________________________________________________
conv2d_13 (Conv2D) (None, 64, 64, 256) 524544 up_sampling2d_1[0][0]
__________________________________________________________________________________________________
concatenate_1 (Concatenate) (None, 64, 64, 512) 0 batch_normalization_5[0][0]
conv2d_13[0][0]
__________________________________________________________________________________________________
conv2d_14 (Conv2D) (None, 64, 64, 256) 1179904 concatenate_1[0][0]
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 64, 64, 256) 1024 conv2d_14[0][0]
__________________________________________________________________________________________________
conv2d_15 (Conv2D) (None, 64, 64, 256) 590080 batch_normalization_12[0][0]
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 64, 64, 256) 1024 conv2d_15[0][0]
__________________________________________________________________________________________________
up_sampling2d_2 (UpSampling2D) (None, 128, 128, 256 0 batch_normalization_13[0][0]
__________________________________________________________________________________________________
conv2d_16 (Conv2D) (None, 128, 128, 128 131200 up_sampling2d_2[0][0]
__________________________________________________________________________________________________
concatenate_2 (Concatenate) (None, 128, 128, 256 0 batch_normalization_3[0][0]
conv2d_16[0][0]
__________________________________________________________________________________________________
conv2d_17 (Conv2D) (None, 128, 128, 128 295040 concatenate_2[0][0]
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 128, 128, 128 512 conv2d_17[0][0]
__________________________________________________________________________________________________
conv2d_18 (Conv2D) (None, 128, 128, 128 147584 batch_normalization_14[0][0]
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 128, 128, 128 512 conv2d_18[0][0]
__________________________________________________________________________________________________
up_sampling2d_3 (UpSampling2D) (None, 256, 256, 128 0 batch_normalization_15[0][0]
__________________________________________________________________________________________________
conv2d_19 (Conv2D) (None, 256, 256, 64) 32832 up_sampling2d_3[0][0]
__________________________________________________________________________________________________
concatenate_3 (Concatenate) (None, 256, 256, 128 0 batch_normalization_1[0][0]
conv2d_19[0][0]
__________________________________________________________________________________________________
conv2d_20 (Conv2D) (None, 256, 256, 64) 73792 concatenate_3[0][0]
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 256, 256, 64) 256 conv2d_20[0][0]
__________________________________________________________________________________________________
conv2d_21 (Conv2D) (None, 256, 256, 64) 36928 batch_normalization_16[0][0]
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 256, 256, 64) 256 conv2d_21[0][0]
__________________________________________________________________________________________________
conv2d_22 (Conv2D) (None, 256, 256, 2) 130 batch_normalization_17[0][0]
==================================================================================================
Total params: 31,054,210
Trainable params: 31,042,434
Non-trainable params: 11,776
__________________________________________________________________________________________________
None
I didn't find any example for binary cross entropy. Ways [1.0,2.0] and {0:1.0,1:2.0} respectively for methods compile and fit get errors and dont work for me.
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|