'2020/05'에 해당되는 글 1건

  1. 2020.05.30 블록과 함께하는 파이썬 딥러닝 케라스

블록과 함께하는 파이썬 딥러닝 케라스

카테고리 없음 2020. 5. 30. 13:01

conda create -n deep python=3.7
conda activate deep
pip install tensorflow==2.0

 

 

 

인공지능 > 머신러닝 > 딥러닝

 

기계학습

 

기존의 코딩

-> 조건을 라인바이라인으로 긴 코드로 써내려 가는 일

 

앞으로의 코딩

-> 조건을 학습 도델의 여러 가중치로 변환하는 일

 

머신러닝

-> 가르쳐주는 학습 -> 분류, 회귀

-> 자율학습 -> 군집

 

지도학습 - KNN ( k Nearest Neighbors ) : 최근접 이웃 알고리즘

 

 

 

 

 

 

 

 

 

 

=======================================

 

import random
import numpy as np

r=[] # 여자 1
b=[] # 남자 0

for i in range(50):
r.append([random.randint(40, 70), random.randint(140, 180), 1])
b.append([random.randint(60, 90), random.randint(160, 200), 0])

def distance(x,y):
#두 점 사이의 거리를 구하는 함수
return np.sqrt(pow((x[0]-y[0]),2) + pow((x[1]-y[1]),2))

def knn(x,y,k):

result=[]
cnt=0

for i in range(len(y)):
result.append([distance(x,y[i]),y[i][2]])
result.sort()

for i in range(k):
if(result[i][1]==1):
cnt += 1

if (cnt > (k/2)):
print ("당신은 여자입니다.")
else:
print ("당신은 남자입니다.")

weight = input("몸무게를 입력해 주세요. ")
height = input("키를 입력해 주세요. ")
num = input("k를 입력해 주세요. ")

new = [int(weight), int(height)]
knn(new, r+b, int(num))

 

=======================================

 

 

 

 

 

pip install matplotlib

 

 

 

 

 

import random
import numpy as np

r=[] # 여자 1
b=[] # 남자 0

for i in range(50):
r.append([random.randint(40, 70), random.randint(140, 180), 1])
b.append([random.randint(60, 90), random.randint(160, 200), 0])

def distance(x,y):
#두 점 사이의 거리를 구하는 함수
return np.sqrt(pow((x[0]-y[0]),2) + pow((x[1]-y[1]),2))

def knn(x,y,k):

result=[]
cnt=0

for i in range(len(y)):
result.append([distance(x,y[i]),y[i][2]])
result.sort()

for i in range(k):
if(result[i][1]==1):
cnt += 1

if (cnt > (k/2)):
print ("당신은 여자입니다.")
else:
print ("당신은 남자입니다.")

import matplotlib.pyplot as plt

rr = np.array(r)
bb = np.array(b)
for i,j in rr[:,:2]:
plt.plot(i,j,'or')
for i,j in bb[:,:2]:
plt.plot(i,j,'ob')
plt.show()

weight = input("몸무게를 입력해 주세요. ")
height = input("키를 입력해 주세요. ")
num = input("k를 입력해 주세요. ")

new = [int(weight), int(height)]
knn(new, r+b, int(num))

 

[참고]

 

https://github.com/ahnsugi/hh/blob/master/%EC%98%9B%EB%82%A0%EA%B1%B0/zoostar_kmeans.ipynb

 

 

딥러닝

 

사람의 뇌세포와 가상의 모델 - 인공신경망

 

매우 많은 Hidden Layers = Deep

 

수확가속의 법칙, 무어의 법칙

 

- 합성곱 신경망 (Convolutional Neural Network, CNN)
- 순환 신경망 (Recurrent Neural Network, RNN)

 

- 옵티마이저: Gradient Decent

- 활성화 함수

- Batch 학습

- DropOut

- LearningRate

- 순전파, 역전파

 

- 텐서플로우2.0 케라스

-> 데이터셋 생성

-> 모델구성

-> 모델 학습 과정 설정

-> 모델학습

-> 학습과정 

-> 모델평가

-> 모델사용

 

데이터 셋

-> 훈련셋, 시험셋, 실전

-> 훈련셋, 검증셋, 시험셋, 실전

 

* epochs

* Model.fit()

* EarlyStopping

 

* 성능평가지표: 정확도, 정밀도, 재현율, Accuracy, Precision, Recall

 - A, B 의사 중 A의사는 암환자를 100% 찾음, B의사는 50% 찾음, A의사는 진료한 모든 환자에게 암진단, B의사는 암으로 진단한 환자는 100% 암환자

 

* 퍼셉트론

 - 다수의 신호를 입력으로 받아 하나의 신호를 출력

 - 퍼셉트론 신호는 흐름을 만들고 정보를 앞으로 전달한다.

 - 퍼셉트론 신호는 '흐른다/안 흐른다' 두 가지의 값을 가진다.

 

 

 

 

 

* 다층 퍼셉트론

 

 

# 논리회로 And , OR , NANE

# w1 * x1 + w2 * x2 + b > 0 흐른다.
# w1 * x1 + w2 * x2 + b <= 0 흐르지 않는다.

import numpy as np


def AND(x1, x2):
x = np.array([x1, x2])
w = np.array([0.5, 0.5])
b = -0.7

tmp = np.sum(w * x) + b
if tmp <= 0:
return 0 # 흐르지 않는다.
else:
return 1 # 흐른다.


def OR(x1, x2):
x = np.array([x1, x2])
w = np.array([0.5, 0.5])
b = -0.2

tmp = np.sum(w * x) + b
if tmp <= 0:
return 0 # 흐르지 않는다.
else:
return 1 # 흐른다.


def NAND(x1, x2):
x = np.array([x1, x2])
w = np.array([-0.5, -0.5])
b = 0.7

tmp = np.sum(w * x) + b
if tmp <= 0:
return 0 # 흐르지 않는다.
else:
return 1 # 흐른다.


print("AND")
for i in [(0, 0), (1, 0), (0, 1), (1, 1)]:
y = AND(i[0], i[1])
print(str(i) + " -> " + str(y))
print()
print("OR")
for i in [(0, 0), (1, 0), (0, 1), (1, 1)]:
y = OR(i[0], i[1])
print(str(i) + " -> " + str(y))
print()
print("NAND")
for i in [(0, 0), (1, 0), (0, 1), (1, 1)]:
y = NAND(i[0], i[1])
print(str(i) + " -> " + str(y))

print()
print("XOR")
for i in [(0, 0), (1, 0), (0, 1), (1, 1)]:
s1 = NAND(i[0], i[1])
s2 = OR(i[0], i[1])
y = AND(s1, s2)

print(str(i) + " -> " + str(y))

 

 

* 다층 퍼셉트론 레이어

- 순전파(Feedforward)

- 역전파(Backpropagation)

 

* Dense(8, input_dim=4, init='uniform', activation='relu'), 크고 낮을수록 미치는 영향이 적다.

- 첫번째 인자: 출력 뉴런의 수를 설정

- input_dim: 입력 뉴런의 수를 설정

- init: 가중치 초기화 방법 설정, 'uniform': 균일 분포, 'normal': 가우시안 분포

- activation: 활성화 함수 설정, 'linear', 'relu', 'sigmoid': 시그모이드 함수, 이진 분류 문제에 출력층에 주로 쓰임, 'softmax': 소프트맥스 함수, 다중 클래스 분류 출력층에 주로 쓰임

 

 

 

 

[출처] https://tykimos.github.io/2017/01/27/MLP_Layer_Talk/

 

 

다층 퍼셉트론 레이어 이야기

이번에는 케라스에서 사용되는 레이어(layer, 층) 개념에 대해서 알아봅니다. 케라스의 핵심 데이터 구조는 모델이고, 이 모델을 구성하는 것이 레이어입니다. 간단히 뉴런에 대해서 알아본 다음,

tykimos.github.io

 

https://colab.research.google.com/drive/1HJlWOpqLyvaAIhjPfq37wIYYN-GQblvV

 

Google Colaboratory

 

colab.research.google.com

 

# -*- coding: utf-8 -*-
"""2x_deep_2005.ipynb

Automatically generated by Colaboratory.

Original file is located at
https://colab.research.google.com/drive/1SaUfrHhyDRsbFw_TM8QqMjaeR269MA-i
"""

!pip install tensorflow==2.0

# Commented out IPython magic to ensure Python compatibility.
import tensorflow as tf
# print(tf.__version__)
import tensorflow.keras.utils as utils
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np
import matplotlib.pyplot as plt
# %matplotlib inline

# 1. 데이터셋 준비하기
X_train = np.array(
[
1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9
]
)

Y_train = np.array(
[
2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18
])

X_val = np.array(
[
1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9,1,2,3,4,5,6,7,8,9
])

Y_val = np.array(
[
2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18,2,4,6,8,10,12,14,16,18
])


# 라벨링 전환
Y_train = utils.to_categorical(Y_train,19)
Y_val = utils.to_categorical(Y_val,19)


model = Sequential()
model.add(Dense(units=38, input_dim=1, activation='elu'))
model.add(Dense(units=19, activation='softmax'))

model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])

# 4. 모델 학습시키기
hist = model.fit(X_train, Y_train, epochs=200, batch_size=1, verbose=0, validation_data=(X_val, Y_val))




fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(hist.history['loss'], 'y', label='train loss')
loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')

acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')
acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')

loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax.set_ylabel('accuray')

loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')

plt.show()

# 6. 모델 사용하기
X_test = np.array([
1, 2, 3, 4, 5, 6, 7, 8, 9
])
Y_test = np.array([
[0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1]

])
loss_and_metrics = model.evaluate(X_test, Y_test, batch_size=1)

print('')
print('loss : ' + str(loss_and_metrics[0]))
print('accuray : ' + str(loss_and_metrics[1]))

# Commented out IPython magic to ensure Python compatibility.
import tensorflow as tf
# print(tf.__version__)
import tensorflow.keras.utils as utils
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np
import matplotlib.pyplot as plt
import random
# %matplotlib inline
x_data = []
for i in range(100):
x_data.append([random.randint(40, 60),random.randint(140, 170)])
x_data.append([random.randint(60, 90),random.randint(170, 200)])
y_data = []
for i in range(100):
y_data.append(1)#
y_data.append(0)#
# 1. 데이터셋 준비하기
X_train = np.array([x_data])
X_train = X_train.reshape(200,2)

Y_train = np.array(y_data)
Y_train = Y_train.reshape(200,)



print(X_train[0])
print(Y_val[0])

model = Sequential()
model.add(Dense(20, input_dim=2, activation='relu'))
model.add(Dense(10, activation='relu'))
model.add(Dense(1, activation='sigmoid'))

model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])

# 4. 모델 학습시키기
hist = model.fit(X_train, Y_train, epochs=200, batch_size=10, verbose=1)


fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(hist.history['loss'], 'y', label='train loss')
# loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')

acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')
# acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')

loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax.set_ylabel('accuray')

loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')

plt.show()

x_test = np.array([[50,150],[80,180],[75,170],[60,150],[45,155]])
x_test = x_test.reshape(5,2)
y_test = np.array([1,0,0,1,1])
scores = model.evaluate(x_test, y_test)
print("%s: %.2f%%" %(model.metrics_names[1], scores[1]*100))

x_data = []
for i in range(20):
x_data.append([random.randint(40, 60),random.randint(140, 170)])
x_data.append([random.randint(60, 90),random.randint(170, 200)])
y_data = []
for i in range(20):
y_data.append(1)#
y_data.append(0)#
# 1. 데이터셋 준비하기
x_test = np.array([x_data])
x_test = x_test.reshape(40,2)

y_test = np.array(y_data)
y_test = y_test.reshape(40,)

yhat = model.predict_classes(x_test)
for i in range(len(x_test)):
print(x_test[i])
print('True : ' + str(y_test[i]) + ', Predict : ' + str(yhat[i]))
print()

 

============= result ===============

 

9/1 [==============================================================================================================================================================================================================================================================================] - 0s 7ms/sample - loss: 0.0889 - accuracy: 1.0000

loss : 0.1424690584341685
accuray : 1.0
[ 58 149]
[0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]
Train on 200 samples
Epoch 1/200

 10/200 [>.............................] - ETA: 4s - loss: 16.2614 - accuracy: 0.5000
200/200 [==============================] - 0s 1ms/sample - loss: 6.5254 - accuracy: 0.4700
Epoch 2/200

 10/200 [>.............................] - ETA: 0s - loss: 4.5406 - accuracy: 0.2000
200/200 [==============================] - 0s 156us/sample - loss: 1.0952 - accuracy: 0.5850
Epoch 3/200

 10/200 [>.............................] - ETA: 0s - loss: 0.7697 - accuracy: 0.5000
200/200 [==============================] - 0s 78us/sample - loss: 0.6927 - accuracy: 0.5600
Epoch 4/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3611 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.5439 - accuracy: 0.7400
Epoch 5/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5720 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.5198 - accuracy: 0.7550
Epoch 6/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4730 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.5582 - accuracy: 0.6700
Epoch 7/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4560 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.5506 - accuracy: 0.6900
Epoch 8/200

 10/200 [>.............................] - ETA: 0s - loss: 0.7339 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.4822 - accuracy: 0.7800
Epoch 9/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4067 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.4541 - accuracy: 0.7600
Epoch 10/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4785 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4596 - accuracy: 0.7500
Epoch 11/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4516 - accuracy: 0.6000
200/200 [==============================] - 0s 156us/sample - loss: 0.4577 - accuracy: 0.7450
Epoch 12/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3985 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4776 - accuracy: 0.7250
Epoch 13/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3117 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.4333 - accuracy: 0.7750
Epoch 14/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4142 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4250 - accuracy: 0.8100
Epoch 15/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6721 - accuracy: 0.5000
200/200 [==============================] - 0s 78us/sample - loss: 0.4441 - accuracy: 0.7500
Epoch 16/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4186 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4562 - accuracy: 0.7650
Epoch 17/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4121 - accuracy: 0.9000
200/200 [==============================] - 0s 156us/sample - loss: 0.5091 - accuracy: 0.7350
Epoch 18/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4768 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4194 - accuracy: 0.7650
Epoch 19/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4975 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4742 - accuracy: 0.7650
Epoch 20/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3676 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4718 - accuracy: 0.7550
Epoch 21/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4497 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.5758 - accuracy: 0.7350
Epoch 22/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4378 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4561 - accuracy: 0.7700
Epoch 23/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3829 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4815 - accuracy: 0.7200
Epoch 24/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4210 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4250 - accuracy: 0.7700
Epoch 25/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5895 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4560 - accuracy: 0.7500
Epoch 26/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4725 - accuracy: 0.6000
200/200 [==============================] - 0s 156us/sample - loss: 0.4468 - accuracy: 0.7400
Epoch 27/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4354 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4045 - accuracy: 0.7500
Epoch 28/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3011 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4104 - accuracy: 0.7450
Epoch 29/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4523 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4062 - accuracy: 0.7900
Epoch 30/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3128 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4347 - accuracy: 0.7850
Epoch 31/200

 10/200 [>.............................] - ETA: 0s - loss: 0.0796 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.4021 - accuracy: 0.8100
Epoch 32/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4472 - accuracy: 0.7000
200/200 [==============================] - 0s 156us/sample - loss: 0.4435 - accuracy: 0.7250
Epoch 33/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4080 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4114 - accuracy: 0.7450
Epoch 34/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5712 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.4192 - accuracy: 0.7250
Epoch 35/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3591 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.5734 - accuracy: 0.7250
Epoch 36/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3946 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.5369 - accuracy: 0.7250
Epoch 37/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1608 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.4410 - accuracy: 0.7700
Epoch 38/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3579 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4392 - accuracy: 0.7500
Epoch 39/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2949 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4363 - accuracy: 0.7400
Epoch 40/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6004 - accuracy: 0.6000
200/200 [==============================] - 0s 156us/sample - loss: 0.4195 - accuracy: 0.7450
Epoch 41/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2256 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4542 - accuracy: 0.7250
Epoch 42/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3459 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4617 - accuracy: 0.7400
Epoch 43/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5157 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.4511 - accuracy: 0.7450
Epoch 44/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3218 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4178 - accuracy: 0.7750
Epoch 45/200

 10/200 [>.............................] - ETA: 0s - loss: 0.7861 - accuracy: 0.5000
200/200 [==============================] - 0s 78us/sample - loss: 0.4156 - accuracy: 0.7550
Epoch 46/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4392 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4629 - accuracy: 0.7750
Epoch 47/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1316 - accuracy: 1.0000
200/200 [==============================] - 0s 156us/sample - loss: 0.4117 - accuracy: 0.7650
Epoch 48/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2055 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.4490 - accuracy: 0.7800
Epoch 49/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2155 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.3967 - accuracy: 0.7750
Epoch 50/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4170 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4267 - accuracy: 0.7850
Epoch 51/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4802 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4061 - accuracy: 0.7800
Epoch 52/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5134 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4156 - accuracy: 0.7900
Epoch 53/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6651 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.5397 - accuracy: 0.7450
Epoch 54/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3463 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3947 - accuracy: 0.7750
Epoch 55/200

 10/200 [>.............................] - ETA: 0s - loss: 0.9660 - accuracy: 0.5000
200/200 [==============================] - 0s 78us/sample - loss: 0.4489 - accuracy: 0.7550
Epoch 56/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4397 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4020 - accuracy: 0.7650
Epoch 57/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3729 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4041 - accuracy: 0.7450
Epoch 58/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6306 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4287 - accuracy: 0.7550
Epoch 59/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2280 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3826 - accuracy: 0.7600
Epoch 60/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6788 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4485 - accuracy: 0.7500
Epoch 61/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3731 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3917 - accuracy: 0.7750
Epoch 62/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1518 - accuracy: 1.0000
200/200 [==============================] - 0s 156us/sample - loss: 0.3960 - accuracy: 0.8150
Epoch 63/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6018 - accuracy: 0.7000
200/200 [==============================] - 0s 96us/sample - loss: 0.4171 - accuracy: 0.7800
Epoch 64/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3930 - accuracy: 0.7000
200/200 [==============================] - 0s 98us/sample - loss: 0.3918 - accuracy: 0.7450
Epoch 65/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5992 - accuracy: 0.6000
200/200 [==============================] - 0s 105us/sample - loss: 0.3962 - accuracy: 0.7800
Epoch 66/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4175 - accuracy: 0.7000
200/200 [==============================] - 0s 90us/sample - loss: 0.4002 - accuracy: 0.7450
Epoch 67/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5844 - accuracy: 0.7000
200/200 [==============================] - 0s 90us/sample - loss: 0.3837 - accuracy: 0.7650
Epoch 68/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5052 - accuracy: 0.7000
200/200 [==============================] - 0s 89us/sample - loss: 0.4157 - accuracy: 0.7450
Epoch 69/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3956 - accuracy: 0.9000
200/200 [==============================] - 0s 90us/sample - loss: 0.4248 - accuracy: 0.7800
Epoch 70/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5199 - accuracy: 0.8000
200/200 [==============================] - 0s 85us/sample - loss: 0.4200 - accuracy: 0.7900
Epoch 71/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4047 - accuracy: 0.9000
200/200 [==============================] - 0s 85us/sample - loss: 0.5100 - accuracy: 0.7650
Epoch 72/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3277 - accuracy: 0.8000
200/200 [==============================] - 0s 85us/sample - loss: 0.4040 - accuracy: 0.7800
Epoch 73/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6739 - accuracy: 0.6000
200/200 [==============================] - 0s 95us/sample - loss: 0.4231 - accuracy: 0.7800
Epoch 74/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3925 - accuracy: 0.8000
200/200 [==============================] - 0s 95us/sample - loss: 0.3871 - accuracy: 0.7750
Epoch 75/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6753 - accuracy: 0.7000
200/200 [==============================] - 0s 100us/sample - loss: 0.4168 - accuracy: 0.7900
Epoch 76/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3912 - accuracy: 0.7000
200/200 [==============================] - 0s 88us/sample - loss: 0.4054 - accuracy: 0.7900
Epoch 77/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6154 - accuracy: 0.7000
200/200 [==============================] - 0s 90us/sample - loss: 0.4428 - accuracy: 0.7750
Epoch 78/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5960 - accuracy: 0.7000
200/200 [==============================] - 0s 13us/sample - loss: 0.3986 - accuracy: 0.7800
Epoch 79/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1654 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.4171 - accuracy: 0.8050
Epoch 80/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1757 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3727 - accuracy: 0.8150
Epoch 81/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3086 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.5277 - accuracy: 0.7500
Epoch 82/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5269 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4370 - accuracy: 0.7500
Epoch 83/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1877 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.3963 - accuracy: 0.7600
Epoch 84/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3203 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4250 - accuracy: 0.8100
Epoch 85/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4221 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3688 - accuracy: 0.8000
Epoch 86/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2577 - accuracy: 0.9000
200/200 [==============================] - 0s 156us/sample - loss: 0.3954 - accuracy: 0.7700
Epoch 87/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4802 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3758 - accuracy: 0.7900
Epoch 88/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3513 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3816 - accuracy: 0.7600
Epoch 89/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2873 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3813 - accuracy: 0.7900
Epoch 90/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5436 - accuracy: 0.5000
200/200 [==============================] - 0s 78us/sample - loss: 0.4172 - accuracy: 0.7450
Epoch 91/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6164 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.3887 - accuracy: 0.7900
Epoch 92/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4552 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3916 - accuracy: 0.8050
Epoch 93/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2803 - accuracy: 0.9000
200/200 [==============================] - 0s 160us/sample - loss: 0.4378 - accuracy: 0.7850
Epoch 94/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3582 - accuracy: 0.9000
200/200 [==============================] - 0s 88us/sample - loss: 0.4374 - accuracy: 0.7800
Epoch 95/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3234 - accuracy: 0.8000
200/200 [==============================] - 0s 90us/sample - loss: 0.3766 - accuracy: 0.7750
Epoch 96/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2603 - accuracy: 0.9000
200/200 [==============================] - 0s 90us/sample - loss: 0.3962 - accuracy: 0.7750
Epoch 97/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1336 - accuracy: 1.0000
200/200 [==============================] - 0s 95us/sample - loss: 0.3738 - accuracy: 0.7900
Epoch 98/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6281 - accuracy: 0.7000
200/200 [==============================] - 0s 110us/sample - loss: 0.3938 - accuracy: 0.7950
Epoch 99/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1798 - accuracy: 0.9000
200/200 [==============================] - 0s 110us/sample - loss: 0.3787 - accuracy: 0.8200
Epoch 100/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3603 - accuracy: 0.8000
200/200 [==============================] - 0s 95us/sample - loss: 0.3979 - accuracy: 0.7900
Epoch 101/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4527 - accuracy: 0.7000
200/200 [==============================] - 0s 101us/sample - loss: 0.3985 - accuracy: 0.7950
Epoch 102/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4397 - accuracy: 0.8000
200/200 [==============================] - 0s 95us/sample - loss: 0.3943 - accuracy: 0.7800
Epoch 103/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6585 - accuracy: 0.6000
200/200 [==============================] - 0s 90us/sample - loss: 0.3965 - accuracy: 0.7850
Epoch 104/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4816 - accuracy: 0.7000
200/200 [==============================] - 0s 85us/sample - loss: 0.4059 - accuracy: 0.7900
Epoch 105/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2570 - accuracy: 0.9000
200/200 [==============================] - 0s 88us/sample - loss: 0.3909 - accuracy: 0.8050
Epoch 106/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6726 - accuracy: 0.7000
200/200 [==============================] - 0s 85us/sample - loss: 0.3651 - accuracy: 0.7850
Epoch 107/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5486 - accuracy: 0.8000
200/200 [==============================] - 0s 35us/sample - loss: 0.3696 - accuracy: 0.8000
Epoch 108/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3855 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4372 - accuracy: 0.7650
Epoch 109/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4137 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3599 - accuracy: 0.8000
Epoch 110/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4072 - accuracy: 0.9000
200/200 [==============================] - 0s 156us/sample - loss: 0.3627 - accuracy: 0.8100
Epoch 111/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5283 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.4651 - accuracy: 0.7550
Epoch 112/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1027 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.4787 - accuracy: 0.7950
Epoch 113/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2348 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4438 - accuracy: 0.7900
Epoch 114/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1851 - accuracy: 1.0000
200/200 [==============================] - 0s 156us/sample - loss: 0.4421 - accuracy: 0.7850
Epoch 115/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2998 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3799 - accuracy: 0.8150
Epoch 116/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4731 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3624 - accuracy: 0.7650
Epoch 117/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3444 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3818 - accuracy: 0.7850
Epoch 118/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4014 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4151 - accuracy: 0.7900
Epoch 119/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6365 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3739 - accuracy: 0.8000
Epoch 120/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4881 - accuracy: 0.8000
200/200 [==============================] - 0s 156us/sample - loss: 0.3993 - accuracy: 0.7800
Epoch 121/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3537 - accuracy: 0.9000
200/200 [==============================] - 0s 98us/sample - loss: 0.3545 - accuracy: 0.8000
Epoch 122/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2226 - accuracy: 1.0000
200/200 [==============================] - 0s 94us/sample - loss: 0.3644 - accuracy: 0.8100
Epoch 123/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1925 - accuracy: 1.0000
200/200 [==============================] - 0s 65us/sample - loss: 0.3464 - accuracy: 0.8300
Epoch 124/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2142 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3404 - accuracy: 0.8250
Epoch 125/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1653 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.3813 - accuracy: 0.8200
Epoch 126/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4300 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.3746 - accuracy: 0.7950
Epoch 127/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4202 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3842 - accuracy: 0.8200
Epoch 128/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4288 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3807 - accuracy: 0.7800
Epoch 129/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3345 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3706 - accuracy: 0.8200
Epoch 130/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3459 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.4689 - accuracy: 0.7900
Epoch 131/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4770 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3749 - accuracy: 0.8000
Epoch 132/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5645 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3732 - accuracy: 0.8200
Epoch 133/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2546 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3828 - accuracy: 0.8000
Epoch 134/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3038 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3620 - accuracy: 0.8050
Epoch 135/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2395 - accuracy: 1.0000
200/200 [==============================] - 0s 156us/sample - loss: 0.3843 - accuracy: 0.8100
Epoch 136/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4037 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3736 - accuracy: 0.7900
Epoch 137/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4462 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3507 - accuracy: 0.8300
Epoch 138/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4180 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3924 - accuracy: 0.8100
Epoch 139/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3979 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3608 - accuracy: 0.7900
Epoch 140/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2292 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3503 - accuracy: 0.8150
Epoch 141/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2339 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.4223 - accuracy: 0.7700
Epoch 142/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5970 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.3493 - accuracy: 0.7950
Epoch 143/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3684 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3637 - accuracy: 0.8000
Epoch 144/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4910 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.3589 - accuracy: 0.8000
Epoch 145/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5816 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.4245 - accuracy: 0.8000
Epoch 146/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5162 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3987 - accuracy: 0.7700
Epoch 147/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1691 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.4628 - accuracy: 0.7900
Epoch 148/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1691 - accuracy: 1.0000
200/200 [==============================] - 0s 156us/sample - loss: 0.5235 - accuracy: 0.7850
Epoch 149/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4264 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3403 - accuracy: 0.8100
Epoch 150/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4933 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3594 - accuracy: 0.7950
Epoch 151/200

 10/200 [>.............................] - ETA: 0s - loss: 0.7476 - accuracy: 0.6000
200/200 [==============================] - 0s 78us/sample - loss: 0.3892 - accuracy: 0.8150
Epoch 152/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1152 - accuracy: 1.0000
200/200 [==============================] - 0s 132us/sample - loss: 0.3603 - accuracy: 0.8150
Epoch 153/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6838 - accuracy: 0.7000
200/200 [==============================] - 0s 85us/sample - loss: 0.3950 - accuracy: 0.7750
Epoch 154/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4502 - accuracy: 0.7000
200/200 [==============================] - 0s 85us/sample - loss: 0.3574 - accuracy: 0.8000
Epoch 155/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1410 - accuracy: 1.0000
200/200 [==============================] - 0s 91us/sample - loss: 0.3483 - accuracy: 0.8100
Epoch 156/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1356 - accuracy: 0.9000
200/200 [==============================] - 0s 95us/sample - loss: 0.3412 - accuracy: 0.8200
Epoch 157/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5683 - accuracy: 0.7000
200/200 [==============================] - 0s 95us/sample - loss: 0.4259 - accuracy: 0.7550
Epoch 158/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5531 - accuracy: 0.6000
200/200 [==============================] - 0s 105us/sample - loss: 0.3685 - accuracy: 0.8100
Epoch 159/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4208 - accuracy: 0.8000
200/200 [==============================] - 0s 105us/sample - loss: 0.4319 - accuracy: 0.7650
Epoch 160/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3048 - accuracy: 1.0000
200/200 [==============================] - 0s 105us/sample - loss: 0.3637 - accuracy: 0.8450
Epoch 161/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2311 - accuracy: 0.9000
200/200 [==============================] - 0s 100us/sample - loss: 0.3324 - accuracy: 0.8300
Epoch 162/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3926 - accuracy: 0.9000
200/200 [==============================] - 0s 100us/sample - loss: 0.3592 - accuracy: 0.8200
Epoch 163/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2989 - accuracy: 0.8000
200/200 [==============================] - 0s 95us/sample - loss: 0.3894 - accuracy: 0.8000
Epoch 164/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4270 - accuracy: 0.7000
200/200 [==============================] - 0s 95us/sample - loss: 0.3833 - accuracy: 0.8150
Epoch 165/200

 10/200 [>.............................] - ETA: 0s - loss: 0.8250 - accuracy: 0.7000
200/200 [==============================] - 0s 95us/sample - loss: 0.3792 - accuracy: 0.7900
Epoch 166/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3649 - accuracy: 0.8000
200/200 [==============================] - 0s 95us/sample - loss: 0.3449 - accuracy: 0.8150
Epoch 167/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6517 - accuracy: 0.7000
200/200 [==============================] - 0s 100us/sample - loss: 0.3865 - accuracy: 0.7850
Epoch 168/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4343 - accuracy: 0.7000
200/200 [==============================] - 0s 86us/sample - loss: 0.3665 - accuracy: 0.8350
Epoch 169/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5715 - accuracy: 0.8000
200/200 [==============================] - 0s 65us/sample - loss: 0.3914 - accuracy: 0.8200
Epoch 170/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5215 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3303 - accuracy: 0.8100
Epoch 171/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2657 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3394 - accuracy: 0.7900
Epoch 172/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2558 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3782 - accuracy: 0.8150
Epoch 173/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4426 - accuracy: 0.8000
200/200 [==============================] - 0s 156us/sample - loss: 0.5031 - accuracy: 0.7500
Epoch 174/200

 10/200 [>.............................] - ETA: 0s - loss: 0.1919 - accuracy: 1.0000
200/200 [==============================] - 0s 78us/sample - loss: 0.4066 - accuracy: 0.8300
Epoch 175/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4874 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3562 - accuracy: 0.8050
Epoch 176/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2331 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3431 - accuracy: 0.7900
Epoch 177/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4055 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3542 - accuracy: 0.8150
Epoch 178/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2686 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3564 - accuracy: 0.8000
Epoch 179/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3008 - accuracy: 0.9000
200/200 [==============================] - 0s 78us/sample - loss: 0.3626 - accuracy: 0.7800
Epoch 180/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3370 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3526 - accuracy: 0.8300
Epoch 181/200

 10/200 [>.............................] - ETA: 0s - loss: 0.5098 - accuracy: 0.7000
200/200 [==============================] - 0s 156us/sample - loss: 0.3414 - accuracy: 0.8200
Epoch 182/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4433 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.3270 - accuracy: 0.8100
Epoch 183/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3821 - accuracy: 0.8000
200/200 [==============================] - 0s 106us/sample - loss: 0.3351 - accuracy: 0.8550
Epoch 184/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3687 - accuracy: 0.7000
200/200 [==============================] - 0s 95us/sample - loss: 0.3306 - accuracy: 0.8350
Epoch 185/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3113 - accuracy: 0.9000
200/200 [==============================] - 0s 90us/sample - loss: 0.3409 - accuracy: 0.8300
Epoch 186/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2959 - accuracy: 0.8000
200/200 [==============================] - 0s 40us/sample - loss: 0.3214 - accuracy: 0.8450
Epoch 187/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4543 - accuracy: 0.7000
200/200 [==============================] - 0s 78us/sample - loss: 0.4262 - accuracy: 0.7900
Epoch 188/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4174 - accuracy: 0.8000
200/200 [==============================] - 0s 78us/sample - loss: 0.3230 - accuracy: 0.8250
Epoch 189/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2877 - accuracy: 0.8000
200/200 [==============================] - 0s 143us/sample - loss: 0.3104 - accuracy: 0.8350
Epoch 190/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3349 - accuracy: 0.9000
200/200 [==============================] - 0s 90us/sample - loss: 0.3351 - accuracy: 0.8350
Epoch 191/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3502 - accuracy: 0.9000
200/200 [==============================] - 0s 95us/sample - loss: 0.4169 - accuracy: 0.8150
Epoch 192/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2316 - accuracy: 0.8000
200/200 [==============================] - 0s 105us/sample - loss: 0.5093 - accuracy: 0.7600
Epoch 193/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6008 - accuracy: 0.6000
200/200 [==============================] - 0s 95us/sample - loss: 0.4614 - accuracy: 0.7700
Epoch 194/200

 10/200 [>.............................] - ETA: 0s - loss: 0.4760 - accuracy: 0.8000
200/200 [==============================] - 0s 98us/sample - loss: 0.3435 - accuracy: 0.8100
Epoch 195/200

 10/200 [>.............................] - ETA: 0s - loss: 0.2994 - accuracy: 0.9000
200/200 [==============================] - 0s 90us/sample - loss: 0.3585 - accuracy: 0.8150
Epoch 196/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3662 - accuracy: 0.7000
200/200 [==============================] - 0s 100us/sample - loss: 0.3696 - accuracy: 0.8150
Epoch 197/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3797 - accuracy: 0.8000
200/200 [==============================] - 0s 90us/sample - loss: 0.3721 - accuracy: 0.7800
Epoch 198/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3320 - accuracy: 0.8000
200/200 [==============================] - 0s 89us/sample - loss: 0.3206 - accuracy: 0.8300
Epoch 199/200

 10/200 [>.............................] - ETA: 0s - loss: 0.6940 - accuracy: 0.6000
200/200 [==============================] - 0s 92us/sample - loss: 0.3806 - accuracy: 0.8200
Epoch 200/200

 10/200 [>.............................] - ETA: 0s - loss: 0.3054 - accuracy: 0.9000
200/200 [==============================] - 0s 85us/sample - loss: 0.3535 - accuracy: 0.8000

5/1 [======================================================================================================================================================] - 0s 9ms/sample - loss: 0.5338 - accuracy: 0.8000
accuracy: 80.00%
[ 60 150]
True : 1, Predict : [0]

[ 64 199]
True : 0, Predict : [0]

[ 56 156]
True : 1, Predict : [0]

[ 85 172]
True : 0, Predict : [0]

[ 45 161]
True : 1, Predict : [1]

[ 71 193]
True : 0, Predict : [0]

[ 43 141]
True : 1, Predict : [1]

[ 86 177]
True : 0, Predict : [0]

[ 58 157]
True : 1, Predict : [0]

[ 61 188]
True : 0, Predict : [0]

[ 47 145]
True : 1, Predict : [1]

[ 75 174]
True : 0, Predict : [0]

[ 46 144]
True : 1, Predict : [1]

[ 66 195]
True : 0, Predict : [0]

[ 46 154]
True : 1, Predict : [1]

[ 75 177]
True : 0, Predict : [0]

[ 48 169]
True : 1, Predict : [1]

[ 73 196]
True : 0, Predict : [0]

[ 59 170]
True : 1, Predict : [0]

[ 62 191]
True : 0, Predict : [0]

[ 51 150]
True : 1, Predict : [1]

[ 68 173]
True : 0, Predict : [0]

[ 45 145]
True : 1, Predict : [1]

[ 87 185]
True : 0, Predict : [0]

[ 50 162]
True : 1, Predict : [1]

[ 60 193]
True : 0, Predict : [1]

[ 49 155]
True : 1, Predict : [1]

[ 65 186]
True : 0, Predict : [0]

[ 44 153]
True : 1, Predict : [1]

[ 68 180]
True : 0, Predict : [0]

[ 41 165]
True : 1, Predict : [1]

[ 65 181]
True : 0, Predict : [0]

[ 43 162]
True : 1, Predict : [1]

[ 66 185]
True : 0, Predict : [0]

[ 45 165]
True : 1, Predict : [1]

[ 66 196]
True : 0, Predict : [0]

[ 41 147]
True : 1, Predict : [1]

[ 75 186]
True : 0, Predict : [0]

[ 46 143]
True : 1, Predict : [1]

[ 69 183]
True : 0, Predict : [0]


Process finished with exit code 0

 

 

==========================================

 

https://github.com/HeoJooSeong/zoostar/

 

HeoJooSeong/zoostar

Contribute to HeoJooSeong/zoostar development by creating an account on GitHub.

github.com

 

import tensorflow.keras.utils as utils
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np
#%matplotlib inline
import matplotlib.pyplot as plt

np.random.seed(3)

# 1. 데이터셋 준비하기

# 훈련셋과 시험셋 로딩
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()

# 훈련셋과 검증셋 분리
X_val = X_train[50000:]
Y_val = Y_train[50000:]
X_train = X_train[:50000]
Y_train = Y_train[:50000]

X_train = X_train.reshape(50000, 784).astype('float32') / 255.0
X_val = X_val.reshape(10000, 784).astype('float32') / 255.0
X_test = X_test.reshape(10000, 784).astype('float32') / 255.0

# 훈련셋, 검증셋 고르기
train_rand_idxs = np.random.choice(50000, 700)
val_rand_idxs = np.random.choice(10000, 300)

X_train = X_train[train_rand_idxs]
Y_train = Y_train[train_rand_idxs]
X_val = X_val[val_rand_idxs]
Y_val = Y_val[val_rand_idxs]

# 라벨링 전환
Y_train = utils.to_categorical(Y_train)
Y_val = utils.to_categorical(Y_val)
Y_test = utils.to_categorical(Y_test)

# 2. 모델 구성하기
model = Sequential()
model.add(Dense(units=2, input_dim=28*28, activation='relu'))
model.add(Dense(units=10, activation='softmax'))

# # 3. 모델 엮기
model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])

# # 4. 모델 학습시키기
hist = model.fit(X_train, Y_train, epochs=100, batch_size=10, validation_data=(X_val, Y_val))

fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(hist.history['loss'], 'y', label='train loss')
loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')

acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')
acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')

loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax.set_ylabel('accuray')

loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')

plt.show()

 

# 산 내려오는 작은 오솔길 잘찾기의 발달 계보

 

 

 

 

import tensorflow.keras.utils as utils
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np
#%matplotlib inline
import matplotlib.pyplot as plt
from numpy import argmax

np.random.seed(3)

# 1. 데이터셋 준비하기

# 훈련셋과 시험셋 로딩
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()

# 훈련셋과 검증셋 분리
X_val = X_train[50000:]
Y_val = Y_train[50000:]
X_train = X_train[:50000]
Y_train = Y_train[:50000]

X_train = X_train.reshape(50000, 784).astype('float32') / 255.0
X_val = X_val.reshape(10000, 784).astype('float32') / 255.0
X_test = X_test.reshape(10000, 784).astype('float32') / 255.0

# # 훈련셋, 검증셋 고르기
# train_rand_idxs = np.random.choice(50000, 700)
# val_rand_idxs = np.random.choice(10000, 300)
#
# X_train = X_train[train_rand_idxs]
# Y_train = Y_train[train_rand_idxs]
# X_val = X_val[val_rand_idxs]
# Y_val = Y_val[val_rand_idxs]

# 훈련셋과 검증셋 분리
x_val = X_val[:42000] # 훈련셋의 30%를 검증셋으로 사용
x_train = Y_val[42000:]
y_val = Y_train[:42000] # 훈련셋의 30%를 검증셋으로 사용
y_train = Y_train[42000:]


# 라벨링 전환
Y_train = utils.to_categorical(Y_train)
Y_val = utils.to_categorical(Y_val)
Y_test = utils.to_categorical(Y_test)

# 2. 모델 구성하기
model = Sequential()
model.add(Dense(units=64, input_dim=28*28, activation='relu'))
model.add(Dense(units=10, activation='softmax'))

# # 3. 모델 엮기
model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])

# # 4. 모델 학습시키기
hist = model.fit(X_train, Y_train, epochs=5, batch_size=32, validation_data=(X_val, Y_val))

fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(hist.history['loss'], 'y', label='train loss')
loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')

acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')
acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')

loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax.set_ylabel('accuray')

loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')

plt.show()

# 5. 모델 평가하기
loss_and_metrics = model.evaluate(X_test, Y_test, batch_size=32)
print('')
print('loss_and_metrics : ' + str(loss_and_metrics))

# 6. 모델 사용하기
xhat_idx = np.random.choice(X_test.shape[0], 5)
xhat = X_test[xhat_idx]
yhat = model.predict_classes(xhat)

for i in range(5):
print('True : ' + str(argmax(Y_test[xhat_idx[i]])) + ', Predict : ' + str(yhat[i]))

 

 

* 피마족 인디언 당뇨병 발병 데이터셋

 

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np

# 1. 데이터셋 준비하기
dataset = np.loadtxt("pima-indians-diabetes.csv",delimiter=",")

x_train = dataset[:700,0:8]
y_train = dataset[:700,8]
x_test = dataset[700:,0:8]
y_test = dataset[700:,8]

model = Sequential()
model.add(Dense(12, input_dim=8, activation='relu'))
model.add(Dense(8, activation='relu'))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
hist=model.fit(x_train, y_train, epochs=1500, batch_size=64)

scores = model.evaluate(x_test, y_test)
print("%s: %.2f%%" %(model.metrics_names[1], scores[1]*100))

 

 

import tensorflow as tf
import tensorflow.keras.utils as utils
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np
import matplotlib.pyplot as plt
#%matplotlib inline

np.random.seed(3)

# 1. 데이터셋 준비하기

# 훈련셋과 시험셋 로딩
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()

# 훈련셋과 검증셋 분리
X_val = X_train[50000:]
Y_val = Y_train[50000:]
X_train = X_train[:50000]
Y_train = Y_train[:50000]

X_train = X_train.reshape(50000, 784).astype('float32') / 255.0
X_val = X_val.reshape(10000, 784).astype('float32') / 255.0
X_test = X_test.reshape(10000, 784).astype('float32') / 255.0

# 훈련셋, 검증셋 고르기
train_rand_idxs = np.random.choice(50000, 700)
val_rand_idxs = np.random.choice(10000, 300)

X_train = X_train[train_rand_idxs]
Y_train = Y_train[train_rand_idxs]
X_val = X_val[val_rand_idxs]
Y_val = Y_val[val_rand_idxs]

# 라벨링 전환
Y_train = utils.to_categorical(Y_train)
Y_val = utils.to_categorical(Y_val)
Y_test = utils.to_categorical(Y_test)

# 2. 모델 구성하기
model = Sequential()
model.add(Dense(units=2, input_dim=28*28, activation='relu'))
model.add(Dense(units=10, activation='softmax'))

# 3. 모델 엮기
model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])

# 4. 모델 학습시키기
from tensorflow.keras.callbacks import EarlyStopping
early_stopping = EarlyStopping() # 조기종료 콜백함수 정의
hist = model.fit(X_train, Y_train, epochs=1000, batch_size=10, verbose=0, validation_data=(X_val, Y_val), callbacks=[early_stopping])



fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(hist.history['loss'], 'y', label='train loss')
loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')

acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')
acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')

loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax.set_ylabel('accuray')

loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')

plt.show()

# 6. 모델 사용하기
loss_and_metrics = model.evaluate(X_test, Y_test, batch_size=32)

print('')
print('loss : ' + str(loss_and_metrics[0]))
print('accuray : ' + str(loss_and_metrics[1]))

# 6. 모델 저장하기
from tensorflow.keras.models import load_model
model.save('mnist_mlp_model.h5')

# 1. 실무에 사용할 데이터 준비하기
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_test = x_test.reshape(10000, 784).astype('float32') / 255.0
y_test = utils.to_categorical(y_test)
xhat_idx = np.random.choice(x_test.shape[0], 5)
xhat = x_test[xhat_idx]

# 2. 모델 불러오기
from tensorflow.keras.models import load_model
model = load_model('mnist_mlp_model.h5')

# 3. 모델 사용하기
yhat = model.predict_classes(xhat)

for i in range(5):
print('True : ' + str(np.argmax(y_test[xhat_idx[i]])) + ', Predict : ' + str(yhat[i]))

 

======================================================

 

#### 콜백함수

import tensorflow as tf
import tensorflow.keras.utils as utils
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np
import matplotlib.pyplot as plt
# %matplotlib inline

class CustomHistory(tf.keras.callbacks.Callback):
def init(self):
self.epoch = 0
self.train_loss = []
self.val_loss = []
self.train_acc = []
self.val_acc = []

def on_epoch_end(self, batch, logs={}):
self.train_loss.append(logs.get('loss'))
self.val_loss.append(logs.get('val_loss'))
self.train_acc.append(logs.get('acc'))
self.val_acc.append(logs.get('val_acc'))
if self.epoch % 100 == 0:
print("epoch: {0} - loss: {1:8.6f}".format(self.epoch, logs.get('loss')))

self.epoch += 1


(X_train, Y_train), (X_test, Y_test) = mnist.load_data()

# 훈련셋과 검증셋 분리
X_val = X_train[50000:]
Y_val = Y_train[50000:]
X_train = X_train[:50000]
Y_train = Y_train[:50000]

X_train = X_train.reshape(50000, 784).astype('float32') / 255.0
X_val = X_val.reshape(10000, 784).astype('float32') / 255.0
X_test = X_test.reshape(10000, 784).astype('float32') / 255.0

# 훈련셋, 검증셋 고르기
train_rand_idxs = np.random.choice(50000, 700)
val_rand_idxs = np.random.choice(10000, 300)

X_train = X_train[train_rand_idxs]
Y_train = Y_train[train_rand_idxs]
X_val = X_val[val_rand_idxs]
Y_val = Y_val[val_rand_idxs]

# 라벨링 전환
Y_train = utils.to_categorical(Y_train)
Y_val = utils.to_categorical(Y_val)
Y_test = utils.to_categorical(Y_test)

# 2. 모델 구성하기
model = Sequential()
model.add(Dense(units=2, input_dim=28 * 28, activation='relu'))
model.add(Dense(units=10, activation='softmax'))

# 3. 모델 엮기
model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])

# 4. 모델 학습시키기

custom_hist = CustomHistory()
custom_hist.init()

for epoch_idx in range(1000):
# print ('epochs : ' + str(epoch_idx) )
model.fit(X_train, Y_train, epochs=1, batch_size=10, verbose=0, validation_data=(X_val, Y_val),
callbacks=[custom_hist])

fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(custom_hist.train_loss, 'y', label='train loss')
loss_ax.plot(custom_hist.val_loss, 'r', label='val loss')


loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')

loss_ax.legend(loc='upper left')

plt.show()

# 6. 모델 사용하기
loss_and_metrics = model.evaluate(X_test, Y_test, batch_size=32)

print('')
print('loss : ' + str(loss_and_metrics[0]))
print('accuray : ' + str(loss_and_metrics[1]))

 

============================================================

 

 

### img

pip install pillow
pip install SciPy

 

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
from tensorflow.keras.layers import Flatten
from tensorflow.keras.layers import Conv2D
from tensorflow.keras.layers import MaxPooling2D
from tensorflow.keras.preprocessing.image import ImageDataGenerator

# 랜덤시드 고정시키기
np.random.seed(3)

# 1. 데이터 생성하기
train_datagen = ImageDataGenerator(rescale=1./255)

train_generator = train_datagen.flow_from_directory(
'handwriting_shape/train',
target_size=(24, 24),
batch_size=3,
class_mode='categorical')

test_datagen = ImageDataGenerator(rescale=1./255)

test_generator = test_datagen.flow_from_directory(
'handwriting_shape/test',
target_size=(24, 24),
batch_size=3,
class_mode='categorical')

# 2. 모델 구성하기
model = Sequential()
model.add(Conv2D(32, kernel_size=(3, 3),
activation='relu',
input_shape=(24,24,3)))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dense(3, activation='softmax'))

# 3. 모델 학습과정 설정하기
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# 4. 모델 학습시키기
model.fit_generator(
train_generator,
steps_per_epoch=15,
epochs=50,
validation_data=test_generator,
validation_steps=5)

# 5. 모델 평가하기
print("-- Evaluate --")
scores = model.evaluate_generator(test_generator, steps=5)
print("%s: %.2f%%" %(model.metrics_names[1], scores[1]*100))

# 6. 모델 사용하기
print("-- Predict --")
output = model.predict_generator(test_generator, steps=5)
np.set_printoptions(formatter={'float': lambda x: "{0:0.3f}".format(x)})
print(test_generator.class_indices)
print(output)

 

========================================================

 

# 컨볼루션 신경망 레이어

 

[출처] https://tykimos.github.io/2017/01/27/CNN_Layer_Talk/

 

컨볼루션 신경망 레이어 이야기

이번 강좌에서는 컨볼루션 신경망 모델에서 주로 사용되는 컨볼루션(Convolution) 레이어, 맥스풀링(Max Pooling) 레이어, 플래튼(Flatten) 레이어에 대해서 알아보겠습니다. 각 레이어별로 레이어 구성 �

tykimos.github.io

* 문제 정의하기

- 문제 형태: 다중 클래스 분류

- 입력: 손으로 그린 삼각형, 사각형, 원 이미지

- 출력: 삼각형, 사각형, 원일 확률을 나타내는 벡터

 

* Dropout - 앙상블효과로 성능이 향상 됨

 

* https://www.cs.toronto.edu/~kriz/cifar.html

 

CIFAR-10 and CIFAR-100 datasets

< Back to Alex Krizhevsky's home page The CIFAR-10 and CIFAR-100 are labeled subsets of the 80 million tiny images dataset. They were collected by Alex Krizhevsky, Vinod Nair, and Geoffrey Hinton. The CIFAR-10 dataset The CIFAR-10 dataset consists of 60000

www.cs.toronto.edu

============================================

 

!pip install tensorflow-gpu

 

============================================

 

import tensorflow as tf
import numpy as np
from tensorflow.keras import datasets, layers, models
from matplotlib import pyplot as plt
from tensorflow.keras import layers
from tensorflow.keras import Input
from tensorflow.keras import Model

 

============================================

 

import tensorflow as tf

import numpy as np

from tensorflow.keras.models import Sequential

from tensorflow.keras.layers import Dense

from tensorflow.keras.layers import Flatten

from tensorflow.keras.layers import Conv2D

from tensorflow.keras.layers import MaxPooling2D

from tensorflow.keras.layers import Dropout

 

============================================

 

# keras 데이터 가져오기

cifar10_data = tf.keras.datasets.cifar10.load_data()
((train_data, train_label), (test_data, test_label)) = cifar10_data

 

============================================

 

Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz 170500096/170498071 [==============================] - 6s 0us/step

 

============================================

 

print("train_data_num: {0}, \ntest_data_num: {1}, \ntrain_label_num: {2}, \ntest_label_num: {3},".format(len(train_data), len(test_data), len(train_label), len(test_label)))

 

============================================

 

import tensorflow as tf

print(tf.__version__)
import tensorflow.keras.utils as utils
import numpy as np
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten
from tensorflow.keras.layers import Dropout

width = 28
height = 28

# 1. 데이터셋 생성하기

# 훈련셋과 시험셋 불러오기
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(60000, width, height, 1).astype('float32') / 255.0
x_test = x_test.reshape(10000, width, height, 1).astype('float32') / 255.0

# 훈련셋과 검증셋 분리
x_val = x_train[50000:]
y_val = y_train[50000:]
x_train = x_train[:50000]
y_train = y_train[:50000]

# 데이터셋 전처리 : one-hot 인코딩
y_train = utils.to_categorical(y_train)
y_val = utils.to_categorical(y_val)
y_test = utils.to_categorical(y_test)

# 2. 모델 구성하기
model = Sequential()
model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(width, height, 1)))
model.add(Conv2D(32, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(256, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(10, activation='softmax'))

# 3. 모델 학습과정 설정하기
model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])

# 4. 모델 학습시키기
hist = model.fit(x_train, y_train, epochs=30, batch_size=32, validation_data=(x_val, y_val))

# 5. 학습과정 살펴보기
# % matplotlib
# inline
import matplotlib.pyplot as plt

fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(hist.history['loss'], 'y', label='train loss')
loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')
loss_ax.set_ylim([0.0, 0.5])

acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')
acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')
acc_ax.set_ylim([0.8, 1.0])

loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax.set_ylabel('accuray')

loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')

plt.show()

# 6. 모델 평가하기
loss_and_metrics = model.evaluate(x_test, y_test, batch_size=32)
print('## evaluation loss and_metrics ##')
print(loss_and_metrics)

# 7. 모델 사용하기
yhat_test = model.predict(x_test, batch_size=32)

# % matplotlib
# inline
import matplotlib.pyplot as plt

plt_row = 5
plt_col = 5

plt.rcParams["figure.figsize"] = (10, 10)

f, axarr = plt.subplots(plt_row, plt_col)

cnt = 0
i = 0

while cnt < (plt_row * plt_col):

if np.argmax(y_test[i]) == np.argmax(yhat_test[i]):
i += 1
continue

sub_plt = axarr[int(cnt / plt_row), int(cnt % plt_col)]
sub_plt.axis('off')
sub_plt.imshow(x_test[i].reshape(width, height))
sub_plt_title = 'R: ' + str(np.argmax(y_test[i])) + ' P: ' + str(np.argmax(yhat_test[i]))
sub_plt.set_title(sub_plt_title)

i += 1
cnt += 1

plt.show()

 

============================

 

print(tf.__version__)
import tensorflow.keras.utils as utils
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
import numpy as np

width = 28
height = 28

# 1. 데이터셋 생성하기

# 훈련셋과 시험셋 불러오기
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(60000, width * height).astype('float32') / 255.0
x_test = x_test.reshape(10000, width * height).astype('float32') / 255.0

# 훈련셋과 검증셋 분리
x_val = x_train[50000:]
y_val = y_train[50000:]
x_train = x_train[:50000]
y_train = y_train[:50000]

# 데이터셋 전처리 : one-hot 인코딩
y_train = utils.to_categorical(y_train)
y_val = utils.to_categorical(y_val)
y_test = utils.to_categorical(y_test)

# 2. 모델 구성하기
model = Sequential()
model.add(Dense(256, input_dim=width * height, activation='relu'))
model.add(Dense(256, activation='relu'))
model.add(Dense(256, activation='relu'))
model.add(Dense(256, activation='relu'))
model.add(Dense(10, activation='softmax'))

# 3. 모델 학습과정 설정하기
model.compile(loss='categorical_crossentropy', optimizer='sgd', metrics=['accuracy'])

# 4. 모델 학습시키기
hist = model.fit(x_train, y_train, epochs=50, batch_size=32, validation_data=(x_val, y_val))

# 5. 학습과정 살펴보기
# % matplotlib
# inline
import matplotlib.pyplot as plt

fig, loss_ax = plt.subplots()

acc_ax = loss_ax.twinx()

loss_ax.plot(hist.history['loss'], 'y', label='train loss')
loss_ax.plot(hist.history['val_loss'], 'r', label='val loss')
loss_ax.set_ylim([0.0, 0.5])

acc_ax.plot(hist.history['accuracy'], 'b', label='train acc')
acc_ax.plot(hist.history['val_accuracy'], 'g', label='val acc')
acc_ax.set_ylim([0.8, 1.0])

loss_ax.set_xlabel('epoch')
loss_ax.set_ylabel('loss')
acc_ax.set_ylabel('accuray')

loss_ax.legend(loc='upper left')
acc_ax.legend(loc='lower left')

plt.show()

# 6. 모델 평가하기
loss_and_metrics = model.evaluate(x_test, y_test, batch_size=32)
print('## evaluation loss and_metrics ##')
print(loss_and_metrics)

# 7. 모델 사용하기
yhat_test = model.predict(x_test, batch_size=32)

# % matplotlib
# inline
import matplotlib.pyplot as plt

plt_row = 5
plt_col = 5

plt.rcParams["figure.figsize"] = (10, 10)

f, axarr = plt.subplots(plt_row, plt_col)

cnt = 0
i = 0

while cnt < (plt_row * plt_col):

if np.argmax(y_test[i]) == np.argmax(yhat_test[i]):
i += 1
continue

sub_plt = axarr[int(cnt / plt_row), int(cnt % plt_col)]
sub_plt.axis('off')
sub_plt.imshow(x_test[i].reshape(width, height))
sub_plt_title = 'R: ' + str(np.argmax(y_test[i])) + ' P: ' + str(np.argmax(yhat_test[i]))
sub_plt.set_title(sub_plt_title)

i += 1
cnt += 1

plt.show()

 

================================================

 

https://tykimos.github.io/2017/04/09/RNN_Layer_Talk/

 

순환 신경망 모델 만들어보기

앞서 살펴본 LSTM 레이어를 이용하여 몇가지 순환 신경망 모델을 만들어보고, 각 모델에 “나비야” 동요를 학습시켜보면서 자세히 살펴보겠습니다. 시퀀스 데이터 준비 순환 신경망은 주로 자��

tykimos.github.io

https://musescore.org/ko/download#older-versions

 

Download

Create, play and print beautiful sheet music with the world's most popular notation software

musescore.org

 

import tensorflow as tf

print(tf.__version__)

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, LSTM
import tensorflow.keras.utils as utils
import os

# http://web.mit.edu/music21/doc/usersGuide/usersGuide_08_installingMusicXML.html
import music21

seq = ['g8', 'e8', 'e4', 'f8', 'd8', 'd4', 'c8', 'd8', 'e8', 'f8', 'g8', 'g8', 'g4',
'g8', 'e8', 'e8', 'e8', 'f8', 'd8', 'd4', 'c8', 'e8', 'g8', 'g8', 'e8', 'e8', 'e4',
'd8', 'd8', 'd8', 'd8', 'd8', 'e8', 'f4', 'e8', 'e8', 'e8', 'e8', 'e8', 'f8', 'g4',
'g8', 'e8', 'e4', 'f8', 'd8', 'd4', 'c8', 'e8', 'g8', 'g8', 'e8', 'e8', 'e4']

print("length of seq: {0}".format(len(seq)))

note_seq = ""
for note in seq:
note_seq += note + " "

m = music21.converter.parse("2/4 " + note_seq, format='tinyNotation')

m.show("midi")

m.show()

code2idx = {'c4': 0, 'd4': 1, 'e4': 2, 'f4': 3, 'g4': 4, 'a4': 5, 'b4': 6,
'c8': 7, 'd8': 8, 'e8': 9, 'f8': 10, 'g8': 11, 'a8': 12, 'b8': 13}

idx2code = {0: 'c4', 1: 'd4', 2: 'e4', 3: 'f4', 4: 'g4', 5: 'a4', 6: 'b4',
7: 'c8', 8: 'd8', 9: 'e8', 10: 'f8', 11: 'g8', 12: 'a8', 13: 'b8'}


def seq2dataset(seq, window_size):
dataset = []

for i in range(len(seq) - window_size):
subset = seq[i: (i + window_size + 1)]
dataset.append([code2idx[item] for item in subset])
return np.array(dataset)


test_seq = ['c4', 'd4', 'e4', 'f4', 'g4', 'd8', 'b8']
dataset = seq2dataset(seq=test_seq, window_size=4)
print(dataset)

n_steps = 4
n_inputs = 1

dataset = seq2dataset(seq, window_size=n_steps)

print("dataset.shape: {0}".format(dataset.shape))

x_train = dataset[:, 0: n_steps]
y_train = dataset[:, n_steps]
print("x_train: {0}".format(x_train.shape))
print("y_train: {0}".format(y_train.shape))
print(x_train[0])
print(y_train[0])

max_idx_value = len(code2idx) - 1

print("max_idx_value: {0}".format(max_idx_value))

x_train = x_train / float(max_idx_value)

x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], n_inputs))

y_train = utils.to_categorical(y_train)

one_hot_vec_size = y_train.shape[1]

print("one hot encoding vector size is {0}".format(one_hot_vec_size))
print("After pre-processing")
print("x_train: {0}".format(x_train.shape))
print("y_train: {0}".format(y_train.shape))

model = Sequential()
model.add(LSTM(
units=128,
kernel_initializer='glorot_normal',
bias_initializer='zero',
batch_input_shape=(1, n_steps, n_inputs),
stateful=True
))
model.add(Dense(
units=one_hot_vec_size,
kernel_initializer='glorot_normal',
bias_initializer='zero',
activation='softmax'
))

model.compile(
loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy']
)


class LossHistory(tf.keras.callbacks.Callback):
def init(self):
self.epoch = 0
self.losses = []

def on_epoch_end(self, batch, logs={}):
self.losses.append(logs.get('loss'))

if self.epoch % 100 == 0:
print("epoch: {0} - loss: {1:8.6f}".format(self.epoch, logs.get('loss')))

self.epoch += 1


num_epochs = 1500
history = LossHistory() # 손실 이력 객체 생성

history.init()

for epoch_idx in range(num_epochs + 1):
model.fit(
x=x_train,
y=y_train,
epochs=1,
batch_size=1,
verbose=0,
shuffle=False,
callbacks=[history]
)
if history.losses[-1] < 1e-5:
print("epoch: {0} - loss: {1:8.6f}".format(epoch_idx, history.losses[-1]))
model.reset_states()
break
model.reset_states()

import matplotlib.pyplot as plt
% matplotlib
inline

plt.plot(history.losses)
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train'], loc='upper right')
plt.show()

scores = model.evaluate(x_train, y_train, batch_size=1)
print("{0}: {1}".format(model.metrics_names[1], scores[1] * 100))
model.reset_states()

pred_count = 50 # 최대 예측 개수 정의

# 한 스텝 예측
seq_out = ['g8', 'e8', 'e4', 'f8']
pred_out = model.predict(x_train)

for i in range(pred_count):
idx = np.argmax(pred_out[i]) # one-hot 인코딩을 인덱스 값으로 변환
seq_out.append(idx2code[idx]) # seq_out는 최종 악보이므로 인덱스 값을 코드로 변환하여 저장

model.reset_states()

print("one step prediction : ", seq_out)

seq_in = ['g8', 'c8', 'f4', 'e8']
seq_out = seq_in
seq_in = [code2idx[note] / float(max_idx_value) for note in seq_in] # 코드를 인덱스값으로 변환

for i in range(pred_count):
sample_in = np.array(seq_in)
sample_in = np.reshape(sample_in, (1, n_steps, n_inputs)) # 샘플 수, 타입스텝 수, 속성 수
pred_out = model.predict(sample_in)
idx = np.argmax(pred_out)
seq_out.append(idx2code[idx])
seq_in.append(idx / float(max_idx_value))
seq_in.pop(0)

model.reset_states()

print("full song prediction : ")

for note in seq_out:
print(note, end=" ")

# http://web.mit.edu/music21/doc/usersGuide/usersGuide_08_installingMusicXML.html
import music21

note_seq = ""
for note in seq_out:
note_seq += note + " "

conv_midi = music21.converter.subConverters.ConverterMidi()

m = music21.converter.parse("2/4 " + note_seq, format='tinyNotation')

m.show("midi")

m.show()

m.write("midi", fp="./new_music.mid")

 

 

 

Trackbacks 0 : Comments 0

Write a comment