bro's coding

1. Aanaconda download https://www.anaconda.com/products/individual Individual Edition 🐍 Open Source Anaconda Individual Edition is the world’s most popular Python distribution platform with over 20 million users worldwide. You can trust in our long-term commitment to supporting the Anaconda open-source ecosystem, the platform of choice www.anaconda.com 2. Pycharm download https://www.jetbrains.c..

from keras.layers import Input, Conv2D, Flatten, Dense, Conv2DTranspose, Reshape, Lambda, Activation, BatchNormalization, LeakyReLU, Dropout, ZeroPadding2D, UpSampling2D from keras.layers.merge import _Merge from keras.models import Model, Sequential from keras import backend as K from keras.optimizers import Adam, RMSprop from keras.callbacks import ModelCheckpoint from keras.utils import plot_..

%matplotlib inline import os import matplotlib.pyplot as plt from models.WGANGP import WGANGP from utils.loaders import load_celeb import pickle https://broscoding.tistory.com/330 WGANGP.source from keras.layers import Input, Conv2D, Flatten, Dense, Conv2DTranspose, Reshape, Lambda, Activation, BatchNormalization, LeakyReLU, Dropout, ZeroPadding2D, UpSampling2D from keras.layers.merge impo.. bro..

MAX : 1 MIN : -1 로 정할 경우 [1, 2, 0.5, 3, -1, -3, 4] 는 [1, 1, 0.5, 1, -1, -1, 1] 로 판단

GAN DCGAN 완전층 합성곱

가중치를 -0.01~0.01 로 제한을 해서 진동을 억제 https://broscoding.tistory.com/308 GAN 참고 문헌 http://book.naver.com/bookdb/book_detail.nhn?bid=15660741 미술관에 GAN 딥러닝 실전 프로젝트 창조에 다가서는 GAN의 4가지 생성 프로젝트이 책은 케라스를 사용한 딥러닝 기초부터 AI 분야 최신 알고리즘까지.. broscoding.tistory.com

from keras.layers import Input, Conv2D, Flatten, Dense, Conv2DTranspose, Reshape, Lambda, Activation, BatchNormalization, LeakyReLU, Dropout, ZeroPadding2D, UpSampling2D from keras.layers.merge import _Merge from keras.models import Model, Sequential from keras import backend as K from keras.optimizers import Adam, RMSprop from keras.callbacks import ModelCheckpoint from keras.utils import plot_..

모드 붕괴(mode collapse)는 생성자가 판별자를 속이는 적은 수의 샘플을 찾을 때 일어난다. 한정된 이 샘플 이외에는 다른 샘플을 생성하지 못한다. generator가 0~9(mnist)에서 discriminator 을 잘 속일 수 있는 0과 1만을 만드는 것 mode(모드) : 생성자가 판별자를 항상 속이는 하나의 샘플 GAN 참고 문헌 http://book.naver.com/bookdb/book_detail.nhn?bid=15660741 미술관에 GAN 딥러닝 실전 프로젝트 창조에 다가서는 GAN의 4가지 생성 프로젝트이 책은 케라스를 사용한 딥러닝 기초부터 AI 분야 최신 알고리즘까지.. broscoding.tistory.com

from keras.layers import Input, Conv2D, Flatten, Dense, Conv2DTranspose, Reshape, Lambda, Activation, BatchNormalization, LeakyReLU, Dropout, ZeroPadding2D, UpSampling2D from keras.layers.merge import _Merge from keras.models import Model, Sequential from keras import backend as K from keras.optimizers import Adam, RMSprop from keras.utils import plot_model from keras.initializers import RandomN..

WGAN : wasserstein 손실함수를 GAN에 적용(원래 GAN의 손실 함수는 binary cross entropy) import os import numpy as np import matplotlib.pyplot as plt from models.WGAN import WGAN from utils.loaders import load_cifar # run params SECTION = 'gan' RUN_ID = '0002' DATA_NAME = 'horses' RUN_FOLDER = 'run/{}/'.format(SECTION) RUN_FOLDER += '_'.join([RUN_ID, DATA_NAME]) if not os.path.exists(RUN_FOLDER): os.mkdir(RUN_FOLD..