example.txt

Genetic Algorithm
----->Initializing Population
Training parent_0
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 2s 2ms/step - loss: 2.6259 - acc: 0.1580 - val_loss: 2.3182 - val_acc: 0.1200
Epoch 2/3
1000/1000 [==============================] - 2s 2ms/step - loss: 1.8815 - acc: 0.3380 - val_loss: 1.9496 - val_acc: 0.2900
Epoch 3/3
1000/1000 [==============================] - 2s 2ms/step - loss: 1.5116 - acc: 0.4830 - val_loss: 3.0469 - val_acc: 0.1650
SUMMARY OF parent_0
Model: "sequential_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_1 (Conv2D)            (None, 32, 32, 32)        896
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 30, 30, 32)        9248
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 15, 15, 32)        0
_________________________________________________________________
flatten_1 (Flatten)          (None, 7200)              0
_________________________________________________________________
dense_1 (Dense)              (None, 128)               921728
_________________________________________________________________
dense_2 (Dense)              (None, 10)                1290
=================================================================
Total params: 933,162
Trainable params: 933,162
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  3.0468814468383787

-------------------------------------
Creating individual net_1
-------------------------------------

----->Strong Mutation
Block Mutation
[(0, 0), (1, 2)]
Creating a new block with two Convolutional layers and a Pooling layer
Layer Mutation
Splitting Conv2D layer at index 1
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  32  to  64
Mutating Conv2D layer:
-->changed self.padding from  valid  to  same
Mutating MaxPooling2D layer:
-->changed self.name from  MaxPooling2D  to  AveragePooling2D
Mutating Conv2D layer:
-->changed self.padding from  same  to  valid
Mutating Conv2D layer:
-->changed self.filters from  32  to  16
Mutating Conv2D layer:
-->changed self.filters from  32  to  64
Mutating MaxPooling2D layer:
-->changed self.name from  MaxPooling2D  to  AveragePooling2D
Mutating FullyConnected layer:
-->changed self.units from  128  to  256
Training net_1
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 4s 4ms/step - loss: 2.5362 - acc: 0.1470 - val_loss: 2.2876 - val_acc: 0.0700
Epoch 2/3
1000/1000 [==============================] - 4s 4ms/step - loss: 2.2455 - acc: 0.1670 - val_loss: 2.1790 - val_acc: 0.2050
Epoch 3/3
1000/1000 [==============================] - 4s 4ms/step - loss: 2.0001 - acc: 0.2750 - val_loss: 2.1584 - val_acc: 0.2000
SUMMARY OF net_1
Model: "sequential_2"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_3 (Conv2D)            (None, 32, 32, 64)        1792
_________________________________________________________________
conv2d_4 (Conv2D)            (None, 32, 32, 32)        18464
_________________________________________________________________
average_pooling2d_1 (Average (None, 16, 16, 32)        0
_________________________________________________________________
conv2d_5 (Conv2D)            (None, 14, 14, 64)        18496
_________________________________________________________________
conv2d_6 (Conv2D)            (None, 14, 14, 16)        9232
_________________________________________________________________
conv2d_7 (Conv2D)            (None, 14, 14, 64)        9280
_________________________________________________________________
average_pooling2d_2 (Average (None, 7, 7, 64)          0
_________________________________________________________________
flatten_2 (Flatten)          (None, 3136)              0
_________________________________________________________________
dense_3 (Dense)              (None, 256)               803072
_________________________________________________________________
dense_4 (Dense)              (None, 10)                2570
=================================================================
Total params: 862,906
Trainable params: 862,906
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  2.158364658355713

-------------------------------------
Creating individual net_2
-------------------------------------

----->Strong Mutation
Block Mutation
[(0, 0), (1, 2)]
Creating a new block with two Convolutional layers and a Pooling layer
Layer Mutation
Splitting Conv2D layer at index 1
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  32  to  64
Mutating MaxPooling2D layer:
-->changed self.padding from  same  to  valid
Mutating Conv2D layer:
-->changed self.filters from  64  to  32
Mutating Conv2D layer:
-->changed self.filters from  64  to  128
Mutating FullyConnected layer:
-->changed self.units from  128  to  256
Training net_2
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 9s 9ms/step - loss: 2.9179 - acc: 0.1290 - val_loss: 2.2939 - val_acc: 0.0950
Epoch 2/3
1000/1000 [==============================] - 5s 5ms/step - loss: 2.2981 - acc: 0.1580 - val_loss: 2.4833 - val_acc: 0.0750
Epoch 3/3
1000/1000 [==============================] - 6s 6ms/step - loss: 2.2161 - acc: 0.1990 - val_loss: 2.1947 - val_acc: 0.2250
SUMMARY OF net_2
Model: "sequential_3"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_8 (Conv2D)            (None, 32, 32, 64)        1792
_________________________________________________________________
conv2d_9 (Conv2D)            (None, 30, 30, 32)        18464
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 15, 15, 32)        0
_________________________________________________________________
conv2d_10 (Conv2D)           (None, 15, 15, 128)       36992
_________________________________________________________________
conv2d_11 (Conv2D)           (None, 15, 15, 32)        36896
_________________________________________________________________
conv2d_12 (Conv2D)           (None, 15, 15, 128)       36992
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 8, 8, 128)         0
_________________________________________________________________
flatten_3 (Flatten)          (None, 8192)              0
_________________________________________________________________
dense_5 (Dense)              (None, 256)               2097408
_________________________________________________________________
dense_6 (Dense)              (None, 10)                2570
=================================================================
Total params: 2,231,114
Trainable params: 2,231,114
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  2.1947326183319094

-------------------------------------
Initial Population:
net_1 :  2.158364658355713
net_2 :  2.1947326183319094
parent_0 :  3.0468814468383787
-------------------------------------


-------------------------------------
Generation 1
-------------------------------------

Creating Child 0
----->Elitism selection
Selected net_1 and net_2 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  64  to  128
Child has been mutated
Training net_3
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 6s 6ms/step - loss: 3.1392 - acc: 0.1500 - val_loss: 3.0999 - val_acc: 0.1550
Epoch 2/3
1000/1000 [==============================] - 5s 5ms/step - loss: 2.1084 - acc: 0.2780 - val_loss: 2.2486 - val_acc: 0.1750
Epoch 3/3
1000/1000 [==============================] - 5s 5ms/step - loss: 1.7679 - acc: 0.4150 - val_loss: 2.0289 - val_acc: 0.2900
SUMMARY OF net_3
Model: "sequential_4"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_13 (Conv2D)           (None, 32, 32, 128)       3584
_________________________________________________________________
conv2d_14 (Conv2D)           (None, 30, 30, 32)        36896
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 15, 15, 32)        0
_________________________________________________________________
flatten_4 (Flatten)          (None, 7200)              0
_________________________________________________________________
dense_7 (Dense)              (None, 256)               1843456
_________________________________________________________________
dense_8 (Dense)              (None, 10)                2570
=================================================================
Total params: 1,886,506
Trainable params: 1,886,506
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  2.0289092540740965
----->Evolution: Child net_3 with fitness 2.0289092540740965 replaces parent parent_0 with fitness 3.0468814468383787

Creating Child 1
----->Elitism selection
Selected parent_0 and net_1 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Splitting Conv2D layer at index 2
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  64  to  128
Mutating AveragePooling2D layer:
-->changed self.padding from  same  to  valid
Mutating Conv2D layer:
-->changed self.filters from  16  to  32
Mutating Conv2D layer:
-->changed self.filters from  32  to  16
Mutating AveragePooling2D layer:
-->changed self.name from  AveragePooling2D  to  MaxPooling2D
Mutating FullyConnected layer:
-->changed self.units from  256  to  512
Child has been mutated
Training net_4
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 7s 7ms/step - loss: 2.3426 - acc: 0.1350 - val_loss: 2.3832 - val_acc: 0.1400
Epoch 2/3
1000/1000 [==============================] - 6s 6ms/step - loss: 2.1781 - acc: 0.2090 - val_loss: 2.1141 - val_acc: 0.2450
Epoch 3/3
1000/1000 [==============================] - 6s 6ms/step - loss: 1.9379 - acc: 0.3200 - val_loss: 2.3498 - val_acc: 0.1700
SUMMARY OF net_4
Model: "sequential_5"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_15 (Conv2D)           (None, 32, 32, 128)       3584
_________________________________________________________________
conv2d_16 (Conv2D)           (None, 32, 32, 32)        36896
_________________________________________________________________
average_pooling2d_3 (Average (None, 16, 16, 32)        0
_________________________________________________________________
conv2d_17 (Conv2D)           (None, 14, 14, 64)        18496
_________________________________________________________________
conv2d_18 (Conv2D)           (None, 14, 14, 32)        18464
_________________________________________________________________
conv2d_19 (Conv2D)           (None, 14, 14, 32)        9248
_________________________________________________________________
conv2d_20 (Conv2D)           (None, 14, 14, 16)        4624
_________________________________________________________________
max_pooling2d_5 (MaxPooling2 (None, 7, 7, 16)          0
_________________________________________________________________
flatten_5 (Flatten)          (None, 784)               0
_________________________________________________________________
dense_9 (Dense)              (None, 512)               401920
_________________________________________________________________
dense_10 (Dense)             (None, 10)                5130
=================================================================
Total params: 498,362
Trainable params: 498,362
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  2.3497649002075196
----->Evolution: Child net_4 with fitness 2.3497649002075196 is discarded

-------------------------------------
Generation 2
-------------------------------------

Creating Child 0
----->Tournament selection
Selected parent_0 and net_1 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  64  to  128
Mutating Conv2D layer:
-->changed self.filters from  32  to  64
Child has been mutated
Training net_3
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 9s 9ms/step - loss: 3.1879 - acc: 0.1300 - val_loss: 2.1451 - val_acc: 0.2500
Epoch 2/3
1000/1000 [==============================] - 8s 8ms/step - loss: 2.2669 - acc: 0.2570 - val_loss: 3.1800 - val_acc: 0.1850
Epoch 3/3
1000/1000 [==============================] - 8s 8ms/step - loss: 1.7932 - acc: 0.3640 - val_loss: 2.1017 - val_acc: 0.2950
SUMMARY OF net_3
Model: "sequential_6"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_21 (Conv2D)           (None, 32, 32, 128)       3584
_________________________________________________________________
conv2d_22 (Conv2D)           (None, 32, 32, 64)        73792
_________________________________________________________________
average_pooling2d_4 (Average (None, 16, 16, 64)        0
_________________________________________________________________
flatten_6 (Flatten)          (None, 16384)             0
_________________________________________________________________
dense_11 (Dense)             (None, 256)               4194560
_________________________________________________________________
dense_12 (Dense)             (None, 10)                2570
=================================================================
Total params: 4,274,506
Trainable params: 4,274,506
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  2.101689214706421
----->Evolution: Child net_3 with fitness 2.101689214706421 replaces parent net_2 with fitness 2.1947326183319094

Creating Child 1
----->Tournament selection
Selected parent_0 and parent_0 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Parameters Mutation
Mutating MaxPooling2D layer:
-->changed self.padding from  valid  to  same
Mutating FullyConnected layer:
-->changed self.units from  256  to  512
Child has been mutated
Training net_4
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 7s 7ms/step - loss: 2.8565 - acc: 0.1740 - val_loss: 8.6546 - val_acc: 0.0700
Epoch 2/3
1000/1000 [==============================] - 6s 6ms/step - loss: 2.1402 - acc: 0.3390 - val_loss: 2.4670 - val_acc: 0.2000
Epoch 3/3
1000/1000 [==============================] - 7s 7ms/step - loss: 1.5346 - acc: 0.4700 - val_loss: 2.5484 - val_acc: 0.1800
SUMMARY OF net_4
Model: "sequential_7"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_23 (Conv2D)           (None, 32, 32, 128)       3584
_________________________________________________________________
conv2d_24 (Conv2D)           (None, 30, 30, 32)        36896
_________________________________________________________________
max_pooling2d_6 (MaxPooling2 (None, 15, 15, 32)        0
_________________________________________________________________
flatten_7 (Flatten)          (None, 7200)              0
_________________________________________________________________
dense_13 (Dense)             (None, 512)               3686912
_________________________________________________________________
dense_14 (Dense)             (None, 10)                5130
=================================================================
Total params: 3,732,522
Trainable params: 3,732,522
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  2.5483790826797486
----->Evolution: Child net_4 with fitness 2.5483790826797486 is discarded

-------------------------------------
Generation 3
-------------------------------------

Creating Child 0
----->Proportionate selection
Selected net_2 and net_1 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Splitting Conv2D layer at index 1
Parameters Mutation
Mutating Conv2D layer:
-->changed self.padding from  same  to  valid
Mutating Conv2D layer:
-->changed self.filters from  32  to  64
Mutating Conv2D layer:
-->changed self.padding from  valid  to  same
Mutating Conv2D layer:
-->changed self.padding from  same  to  valid
Mutating FullyConnected layer:
-->changed self.units from  256  to  128.0
Child has been mutated

INDIVIDUAL ABORTED, CREATING A NEW ONE

----->Crossover
Block Mutation
[(0, 0), (1, 2)]
Creating a new block with two Convolutional layers and a Pooling layer
Layer Mutation
Changing Pooling layer at index 0 with Conv2D layer
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  64  to  32
Mutating Conv2D layer:
-->changed self.filters from  32  to  16
Mutating AveragePooling2D layer:
-->changed self.name from  AveragePooling2D  to  MaxPooling2D
Mutating Conv2D layer:
-->changed self.filters from  128  to  256
Mutating Conv2D layer:
-->changed self.padding from  same  to  valid
Training net_3
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 11s 11ms/step - loss: 14.2582 - acc: 0.0910 - val_loss: 14.4257 - val_acc: 0.1050
Epoch 2/3
1000/1000 [==============================] - 20s 20ms/step - loss: 14.5224 - acc: 0.0990 - val_loss: 14.4257 - val_acc: 0.1050
Epoch 3/3
1000/1000 [==============================] - 17s 17ms/step - loss: 14.5224 - acc: 0.0990 - val_loss: 14.4257 - val_acc: 0.1050
SUMMARY OF net_3
Model: "sequential_9"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_31 (Conv2D)           (None, 32, 32, 32)        896
_________________________________________________________________
conv2d_32 (Conv2D)           (None, 32, 32, 16)        4624
_________________________________________________________________
max_pooling2d_7 (MaxPooling2 (None, 16, 16, 16)        0
_________________________________________________________________
conv2d_33 (Conv2D)           (None, 16, 16, 256)       37120
_________________________________________________________________
conv2d_34 (Conv2D)           (None, 16, 16, 256)       590080
_________________________________________________________________
conv2d_35 (Conv2D)           (None, 7, 7, 128)         295040
_________________________________________________________________
flatten_9 (Flatten)          (None, 6272)              0
_________________________________________________________________
dense_16 (Dense)             (None, 256)               1605888
_________________________________________________________________
dense_17 (Dense)             (None, 10)                2570
=================================================================
Total params: 2,536,218
Trainable params: 2,536,218
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  14.42569522857666
----->Evolution: Child net_3 with fitness 14.42569522857666 is discarded

Creating Child 1
----->Proportionate selection
Selected net_1 and net_2 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  128  to  256
Mutating AveragePooling2D layer:
-->changed self.padding from  same  to  valid
Child has been mutated
Training net_4
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 14s 14ms/step - loss: 14.1805 - acc: 0.0940 - val_loss: 14.4257 - val_acc: 0.1050
Epoch 2/3
1000/1000 [==============================] - 12s 12ms/step - loss: 14.5224 - acc: 0.0990 - val_loss: 14.4257 - val_acc: 0.1050
Epoch 3/3
1000/1000 [==============================] - 16s 16ms/step - loss: 14.5224 - acc: 0.0990 - val_loss: 14.4257 - val_acc: 0.1050
SUMMARY OF net_4
Model: "sequential_10"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_36 (Conv2D)           (None, 32, 32, 256)       7168
_________________________________________________________________
conv2d_37 (Conv2D)           (None, 32, 32, 64)        147520
_________________________________________________________________
average_pooling2d_7 (Average (None, 16, 16, 64)        0
_________________________________________________________________
flatten_10 (Flatten)         (None, 16384)             0
_________________________________________________________________
dense_18 (Dense)             (None, 256)               4194560
_________________________________________________________________
dense_19 (Dense)             (None, 10)                2570
=================================================================
Total params: 4,351,818
Trainable params: 4,351,818
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  14.42569522857666
----->Evolution: Child net_4 with fitness 14.42569522857666 is discarded

-------------------------------------
Generation 4
-------------------------------------

Creating Child 0
----->Tournament selection
Selected parent_0 and net_2 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Parameters Mutation
Mutating AveragePooling2D layer:
-->changed self.name from  AveragePooling2D  to  MaxPooling2D
Mutating FullyConnected layer:
-->changed self.units from  256  to  128.0
Child has been mutated

INDIVIDUAL ABORTED, CREATING A NEW ONE

----->Crossover
Block Mutation
[(0, 0), (1, 2)]
Creating a new block with two Convolutional layers and a Pooling layer
Layer Mutation
Changing Pooling layer at index 0 with Conv2D layer
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  128  to  64
Mutating Conv2D layer:
-->changed self.filters from  32  to  16
Mutating MaxPooling2D layer:
-->changed self.padding from  valid  to  same
Mutating Conv2D layer:
-->changed self.filters from  32  to  64
Mutating Conv2D layer:
-->changed self.padding from  same  to  valid
Training net_3
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 6s 6ms/step - loss: 2.9446 - acc: 0.1500 - val_loss: 3.1757 - val_acc: 0.1000
Epoch 2/3
1000/1000 [==============================] - 5s 5ms/step - loss: 2.2292 - acc: 0.1880 - val_loss: 2.0710 - val_acc: 0.2800
Epoch 3/3
1000/1000 [==============================] - 5s 5ms/step - loss: 2.0623 - acc: 0.2890 - val_loss: 1.9787 - val_acc: 0.3000
SUMMARY OF net_3
Model: "sequential_12"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_40 (Conv2D)           (None, 32, 32, 64)        1792
_________________________________________________________________
conv2d_41 (Conv2D)           (None, 30, 30, 16)        9232
_________________________________________________________________
max_pooling2d_9 (MaxPooling2 (None, 15, 15, 16)        0
_________________________________________________________________
conv2d_42 (Conv2D)           (None, 15, 15, 256)       37120
_________________________________________________________________
conv2d_43 (Conv2D)           (None, 15, 15, 64)        147520
_________________________________________________________________
conv2d_44 (Conv2D)           (None, 7, 7, 32)          18464
_________________________________________________________________
flatten_12 (Flatten)         (None, 1568)              0
_________________________________________________________________
dense_21 (Dense)             (None, 256)               401664
_________________________________________________________________
dense_22 (Dense)             (None, 10)                2570
=================================================================
Total params: 618,362
Trainable params: 618,362
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  1.9787270355224609
----->Evolution: Child net_3 with fitness 1.9787270355224609 replaces parent net_1 with fitness 2.158364658355713

Creating Child 1
----->Tournament selection
Selected net_1 and net_2 for reproduction
----->Crossover
Child has been created
----->Soft Mutation
Layer Mutation
Splitting Conv2D layer at index 0
Parameters Mutation
Mutating Conv2D layer:
-->changed self.filters from  64  to  128
Mutating MaxPooling2D layer:
-->changed self.padding from  same  to  valid
Mutating Conv2D layer:
-->changed self.padding from  same  to  valid
Child has been mutated
Training net_4
Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 8s 8ms/step - loss: 2.5167 - acc: 0.1040 - val_loss: 2.2539 - val_acc: 0.1350
Epoch 2/3
1000/1000 [==============================] - 7s 7ms/step - loss: 2.2657 - acc: 0.1800 - val_loss: 2.8756 - val_acc: 0.1600
Epoch 3/3
1000/1000 [==============================] - 7s 7ms/step - loss: 2.1404 - acc: 0.2510 - val_loss: 2.2994 - val_acc: 0.1900
SUMMARY OF net_4
Model: "sequential_13"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_45 (Conv2D)           (None, 32, 32, 128)       3584
_________________________________________________________________
conv2d_46 (Conv2D)           (None, 30, 30, 16)        18448
_________________________________________________________________
max_pooling2d_10 (MaxPooling (None, 15, 15, 16)        0
_________________________________________________________________
conv2d_47 (Conv2D)           (None, 15, 15, 128)       18560
_________________________________________________________________
conv2d_48 (Conv2D)           (None, 15, 15, 128)       147584
_________________________________________________________________
conv2d_49 (Conv2D)           (None, 13, 13, 64)        73792
_________________________________________________________________
conv2d_50 (Conv2D)           (None, 6, 6, 32)          18464
_________________________________________________________________
flatten_13 (Flatten)         (None, 1152)              0
_________________________________________________________________
dense_23 (Dense)             (None, 256)               295168
_________________________________________________________________
dense_24 (Dense)             (None, 10)                2570
=================================================================
Total params: 578,170
Trainable params: 578,170
Non-trainable params: 0
_________________________________________________________________
None
FITNESS:  2.29937292098999
----->Evolution: Child net_4 with fitness 2.29937292098999 is discarded


-------------------------------------
Final Population
-------------------------------------

net_1 :  1.9787270355224609
parent_0 :  2.0289092540740965
net_2 :  2.101689214706421

-------------------------------------
Stats
Best individual at generation 1 has fitness 2.158364658355713 and parameters 862906
Best individual at generation 2 has fitness 2.0289092540740965 and parameters 1886506
Best individual at generation 3 has fitness 2.0289092540740965 and parameters 1886506
Best individual at generation 4 has fitness 2.0289092540740965 and parameters 1886506
Best individual at generation 5 has fitness 1.9787270355224609 and parameters 618362
-------------------------------------

Train on 1000 samples, validate on 200 samples
Epoch 1/3
1000/1000 [==============================] - 6s 6ms/step - loss: 2.6393 - acc: 0.1150 - val_loss: 2.2679 - val_acc: 0.1200
Epoch 2/3
1000/1000 [==============================] - 5s 5ms/step - loss: 2.2652 - acc: 0.1870 - val_loss: 2.9440 - val_acc: 0.1550
Epoch 3/3
1000/1000 [==============================] - 5s 5ms/step - loss: 2.0741 - acc: 0.2450 - val_loss: 3.0250 - val_acc: 0.1850


-------------------------------------
The initial CNN has been evolved successfully in the individual net_1
-------------------------------------



-------------------------------------
Summary of initial CNN
Model: "sequential_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_1 (Conv2D)            (None, 32, 32, 32)        896
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 30, 30, 32)        9248
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 15, 15, 32)        0
_________________________________________________________________
flatten_1 (Flatten)          (None, 7200)              0
_________________________________________________________________
dense_1 (Dense)              (None, 128)               921728
_________________________________________________________________
dense_2 (Dense)              (None, 10)                1290
=================================================================
Total params: 933,162
Trainable params: 933,162
Non-trainable params: 0
_________________________________________________________________
None
Fitness of initial CNN: 3.0468814468383787


-------------------------------------
Summary of evolved individual
Model: "sequential_14"
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
conv2d_51 (Conv2D)           (None, 32, 32, 64)        1792
_________________________________________________________________
conv2d_52 (Conv2D)           (None, 30, 30, 16)        9232
_________________________________________________________________
max_pooling2d_11 (MaxPooling (None, 15, 15, 16)        0
_________________________________________________________________
conv2d_53 (Conv2D)           (None, 15, 15, 256)       37120
_________________________________________________________________
conv2d_54 (Conv2D)           (None, 15, 15, 64)        147520
_________________________________________________________________
conv2d_55 (Conv2D)           (None, 7, 7, 32)          18464
_________________________________________________________________
flatten_14 (Flatten)         (None, 1568)              0
_________________________________________________________________
dense_25 (Dense)             (None, 256)               401664
_________________________________________________________________
dense_26 (Dense)             (None, 10)                2570
=================================================================
Total params: 618,362
Trainable params: 618,362
Non-trainable params: 0
_________________________________________________________________
None
Fitness of the evolved individual: 3.0250408458709717
-------------------------------------