在[之前的文章](https://steemit.com/cn-stem/@hongtao/tensorflow-keras-classification-with-keras)中,我们发现训练组(篮)和验证组(红)的损失函数在20个Epoch之后,向着相反方向变化。训练组损失函数继续下降,验证组损失函数反而在上升,这就是典型的Overfitting(过拟合)现象。  过拟合就是模型过度地学习了训练集的特征,反而没法处理测试集中更一般化的问题。处理过拟合最根本的解决方法当然是获得更多的训练样本。 但是在无法获得更多的训练样本的时候,也有两个最简单的方法,一是对权重进行正则化处理,二就对神经元随机dropout. 关于更多Keras的入门介绍,感兴趣的朋友可以参考[Google的官方教程](https://www.tensorflow.org/tutorials/),更多关于过拟合和欠拟合的相关资料请参考[这里](https://en.wikipedia.org/wiki/Overfitting) 在Keras中我们只需要对模型进行简单改造就能实现正则化和dropout,同样的,为了方便与读者交流,所有的代码都放在了这里: https://github.com/zht007/tensorflow-practice ## L1,L2正则化 模型Overfiting其中一个原因就是某些权重在训练的过程中会被放大,L1正则化相当于给权重加了惩罚因子,从而限制了某些权重过度膨胀。L2相当于对L1惩罚因子乘了个平方,对权重的膨胀加大了惩罚力度。 在Keras的模型中引入L1,或者L2也非常简单,只需要在建立模型的时候加入: ```python kernel_regularizer = keras.regularizers.l1 或 kernel_regularizer = keras.regularizers.l2 ``` 模型如下所示 ```python model = Sequential() model.add(Dense(20,input_shape = (X_train.shape[1],), activation = 'relu', kernel_regularizer = keras.regularizers.l2(0.001))) model.add(Dense(20,input_shape = (X_train.shape[1],), activation = 'relu', kernel_regularizer = keras.regularizers.l2(0.001))) model.add(Dense(10,activation = 'relu', kernel_regularizer = keras.regularizers.l2(0.001))) model.add(Dense(2, activation = 'softmax')) ``` ## Dropout 在需要Dropout的Dense层之后加上: ```python model.add(Dense(2, activation = 'softmax')) ``` 最后我们看看加上L2正则化和Dropout之后的模型是怎么样的。 ```python model = Sequential() model.add(Dense(20,input_shape = (X_train.shape[1],), activation = 'relu', kernel_regularizer = keras.regularizers.l2(0.001))) model.add(keras.layers.Dropout(0.5)) model.add(Dense(20,input_shape = (X_train.shape[1],), activation = 'relu', kernel_regularizer = keras.regularizers.l2(0.001))) model.add(keras.layers.Dropout(0.5)) model.add(Dense(10,activation = 'relu', kernel_regularizer = keras.regularizers.l2(0.001))) model.add(keras.layers.Dropout(0.5)) model.add(Dense(2, activation = 'softmax')) model.summary() ``` Model.summary可以查看整个模型的架构和参数的个数 ``` _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= dense_19 (Dense) (None, 20) 180 _________________________________________________________________ dropout_7 (Dropout) (None, 20) 0 _________________________________________________________________ dense_20 (Dense) (None, 20) 420 _________________________________________________________________ dropout_8 (Dropout) (None, 20) 0 _________________________________________________________________ dense_21 (Dense) (None, 10) 210 _________________________________________________________________ dropout_9 (Dropout) (None, 10) 0 _________________________________________________________________ dense_22 (Dense) (None, 2) 22 ================================================================= Total params: 832 Trainable params: 832 Non-trainable params: 0 _________________________________________________________________ ``` ## 训练结果 最后我们看看正则化和Dropout后端的训练结果吧,是不是比之前漂亮多了。  --- 参考资料和数据来源 https://www.kaggle.com/uciml/pima-indians-diabetes-database https://www.tensorflow.org/tutorials/ https://en.wikipedia.org/wiki/Overfitting --- 同步到我的简书 https://www.jianshu.com/u/bd506afc6fc1
author | hongtao | ||||||
---|---|---|---|---|---|---|---|
permlink | tensorflow-overfitting | ||||||
category | cn-stem | ||||||
json_metadata | {"community":"busy","app":"steemit/0.1","format":"markdown","tags":["cn-stem","tensorflow","keras","busy","cn"],"links":["https://steemit.com/cn-stem/@hongtao/tensorflow-keras-classification-with-keras","https://www.tensorflow.org/tutorials/","https://en.wikipedia.org/wiki/Overfitting","https://github.com/zht007/tensorflow-practice","https://www.kaggle.com/uciml/pima-indians-diabetes-database","https://www.jianshu.com/u/bd506afc6fc1"],"image":["https://ws2.sinaimg.cn/large/006tKfTcgy1g18l4i0ja7j30al0703z2.jpg","https://ws2.sinaimg.cn/large/006tKfTcgy1g18lqpke82j30al0703yx.jpg"]} | ||||||
created | 2019-03-19 17:14:24 | ||||||
last_update | 2019-03-20 10:21:21 | ||||||
depth | 0 | ||||||
children | 10 | ||||||
last_payout | 2019-03-26 17:14:24 | ||||||
cashout_time | 1969-12-31 23:59:59 | ||||||
total_payout_value | 1.940 HBD | ||||||
curator_payout_value | 0.670 HBD | ||||||
pending_payout_value | 0.000 HBD | ||||||
promoted | 0.000 HBD | ||||||
body_length | 3,802 | ||||||
author_reputation | 3,241,267,862,629 | ||||||
root_title | Tensorflow入门——处理overfitting的问题 | ||||||
beneficiaries |
| ||||||
max_accepted_payout | 1,000,000.000 HBD | ||||||
percent_hbd | 10,000 | ||||||
post_id | 81,581,525 | ||||||
net_rshares | 4,194,329,033,010 | ||||||
author_curate_reward | "" |
voter | weight | wgt% | rshares | pct | time |
---|---|---|---|---|---|
wackou | 0 | 78,559,924,930 | 1.29% | ||
tombstone | 0 | 23,000,383,467 | 0.1% | ||
delegate.lafona | 0 | 324,179,909,549 | 20% | ||
drifter1 | 0 | 271,216,524 | 2.16% | ||
lola-carola | 0 | 382,592,708 | 2.16% | ||
kevinwong | 0 | 39,373,089,112 | 0.5% | ||
eric-boucher | 0 | 7,133,051,684 | 2.16% | ||
anwenbaumeister | 0 | 190,207,445 | 4.32% | ||
mammasitta | 0 | 911,654,823 | 0.21% | ||
jesse5th | 0 | 64,805,945 | 50% | ||
arcange | 0 | 29,542,592,702 | 3% | ||
liberosist | 0 | 1,529,039,223 | 4.32% | ||
arconite | 0 | 102,397,023 | 0.25% | ||
raphaelle | 0 | 1,365,348,296 | 3% | ||
ninjace | 0 | 4,543,249,855 | 50% | ||
lemouth | 0 | 48,811,488,975 | 10% | ||
rwilday | 0 | 51,797,230 | 100% | ||
lamouthe | 0 | 11,537,884,070 | 20% | ||
uceph | 0 | 353,711,594 | 100% | ||
justyy | 0 | 46,903,512,803 | 2.05% | ||
whoib | 0 | 524,375,662 | 70% | ||
curie | 0 | 1,368,740,281,621 | 4.32% | ||
hendrikdegrote | 0 | 55,793,111,928 | 4.32% | ||
vact | 0 | 83,301,578,535 | 4.32% | ||
steemstem | 0 | 844,130,157,118 | 20% | ||
gangstayid | 0 | 97,976,023 | 2.16% | ||
busy.org | 0 | 27,390,005 | 0.3% | ||
vodonik | 0 | 80,734,595 | 6.6% | ||
dna-replication | 0 | 5,644,197,583 | 20% | ||
gmedley | 0 | 305,196,917 | 2.16% | ||
diebaasman | 0 | 2,227,819,273 | 12% | ||
moksamol | 0 | 483,283,836 | 2.16% | ||
getrichordie | 0 | 137,070,702 | 2.16% | ||
thatsweeneyguy | 0 | 106,399,958 | 2.16% | ||
szokerobert | 0 | 114,027,078 | 0.86% | ||
bloom | 0 | 82,275,979,583 | 20% | ||
kryzsec | 0 | 846,182,461 | 16% | ||
jiujitsu | 0 | 1,765,832,040 | 2.16% | ||
lekang | 0 | 584,318,307 | 2.16% | ||
helo | 0 | 21,850,373,213 | 10% | ||
samminator | 0 | 6,679,754,253 | 10% | ||
locikll | 0 | 2,794,883,371 | 8.64% | ||
kjaeger | 0 | 54,719,659 | 50% | ||
mahdiyari | 0 | 17,107,134,154 | 10% | ||
lorenzor | 0 | 5,928,660,673 | 50% | ||
aboutyourbiz | 0 | 769,331,674 | 4.32% | ||
alexander.alexis | 0 | 9,957,760,279 | 20% | ||
howtostartablog | 0 | 83,790,296 | 0.43% | ||
suesa | 0 | 109,729,204,556 | 25% | ||
cryptokrieg | 0 | 769,038,243 | 4.32% | ||
rival | 0 | 1,964,711,401 | 2% | ||
slickhustler007 | 0 | 169,321,791 | 2.16% | ||
corsica | 0 | 7,723,317,738 | 20% | ||
makrotheblack | 0 | 72,332,691 | 2.16% | ||
allcapsonezero | 0 | 2,079,632,097 | 2.16% | ||
howo | 0 | 55,984,206,822 | 10% | ||
tsoldovieri | 0 | 1,452,851,283 | 10% | ||
nitego | 0 | 90,537,132 | 1.29% | ||
hotsteam | 0 | 3,203,373,252 | 10% | ||
neumannsalva | 0 | 606,031,886 | 2.16% | ||
wargof | 0 | 200,235,977 | 10% | ||
abigail-dantes | 0 | 329,247,600,940 | 20% | ||
esteemguy | 0 | 153,421,645 | 20% | ||
zonguin | 0 | 1,212,224,586 | 5% | ||
g0nr0gue | 0 | 67,133,848 | 2.16% | ||
alexzicky | 0 | 5,536,255,386 | 5% | ||
mountain.phil28 | 0 | 3,600,893,195 | 25% | ||
jasonbu | 0 | 9,554,229,728 | 25% | ||
coolbuddy | 0 | 0 | 1% | ||
muliadi | 0 | 63,663,390 | 2.16% | ||
tuoficinavirtual | 0 | 84,566,514 | 25% | ||
tanyaschutte | 0 | 81,859,483 | 2% | ||
iamphysical | 0 | 15,865,483,774 | 90% | ||
kingswisdom | 0 | 62,550,754 | 10% | ||
zest | 0 | 3,921,999,611 | 10% | ||
felixrodriguez | 0 | 1,003,382,567 | 10% | ||
azulear | 0 | 496,661,649 | 100% | ||
psicoluigi | 0 | 309,079,275 | 50% | ||
vadimlasca | 0 | 61,212,391 | 4.32% | ||
mr-aaron | 0 | 108,899,895 | 10% | ||
massivevibration | 0 | 3,199,379,940 | 5% | ||
eurodale | 0 | 165,827,945 | 2.16% | ||
reaverza | 0 | 1,158,994,211 | 15% | ||
clweeks | 0 | 208,678,758 | 2.59% | ||
superbing | 0 | 691,035,831 | 7.31% | ||
dokter-purnama | 0 | 256,633,392 | 2.16% | ||
erikkun28 | 0 | 0 | 1% | ||
dailystats | 0 | 2,083,478,472 | 7.31% | ||
jubei333 | 0 | 148,099,709 | 2.16% | ||
jlsplatts | 0 | 221,374,744 | 0.64% | ||
mayowadavid | 0 | 897,779,701 | 10% | ||
poodai | 0 | 135,437,895 | 2.16% | ||
markmorbidity | 0 | 94,895,139 | 2.16% | ||
cryptocurrencyhk | 0 | 526,508,993 | 20% | ||
emdesan | 0 | 140,098,338 | 10% | ||
peaceandwar | 0 | 604,250,979 | 2.16% | ||
enzor | 0 | 245,217,794 | 10% | ||
joendegz | 0 | 74,355,491 | 2.16% | ||
florian-glechner | 0 | 96,185,859 | 0.43% | ||
jesusj1 | 0 | 76,647,336 | 100% | ||
lekosvapenglass | 0 | 70,628,281 | 40% | ||
carloserp-2000 | 0 | 31,305,820,659 | 100% | ||
carlos84 | 0 | 5,744,174,044 | 100% | ||
gra | 0 | 7,744,043,805 | 20% | ||
jianan | 0 | 1,263,587,410 | 7.65% | ||
shayekh2 | 0 | 84,159,743 | 50% | ||
agbona | 0 | 113,243,102 | 5% | ||
ykdesign | 0 | 316,355,569 | 2.16% | ||
pinksteam | 0 | 1,052,864,766 | 10% | ||
aalok | 0 | 106,729,004 | 26% | ||
attoan.cmt | 0 | 578,345,988 | 2.16% | ||
cnbuddy | 0 | 12,363,508,405 | 1% | ||
nicole-st | 0 | 121,389,160 | 2.16% | ||
teukurival | 0 | 162,567,022 | 2.16% | ||
drmake | 0 | 2,326,471,945 | 2.16% | ||
anxin | 0 | 137,259,736 | 7.79% | ||
guga34 | 0 | 495,288,275 | 15% | ||
pechichemena | 0 | 74,021,313 | 0.86% | ||
amestyj | 0 | 2,442,664,477 | 50% | ||
sandracarrascal | 0 | 397,961,019 | 100% | ||
skycae | 0 | 537,325,504 | 4.32% | ||
xanderslee | 0 | 201,446,831 | 4.32% | ||
kenadis | 0 | 4,363,498,119 | 20% | ||
robotics101 | 0 | 2,078,163,889 | 20% | ||
tristan-muller | 0 | 80,198,661 | 20% | ||
gentleshaid | 0 | 8,382,227,077 | 10% | ||
thescubageek | 0 | 239,990,309 | 2.16% | ||
fejiro | 0 | 216,312,225 | 10% | ||
danaedwards | 0 | 369,667,726 | 4.32% | ||
ivymalifred | 0 | 1,773,142,399 | 50% | ||
sco | 0 | 16,840,266,654 | 20% | ||
douglimarbalzan | 0 | 396,177,005 | 100% | ||
ennyta | 0 | 924,804,737 | 50% | ||
hkmoon | 0 | 66,420,711 | 2.16% | ||
rharphelle | 0 | 722,026,409 | 25% | ||
gordon92 | 0 | 176,419,054 | 2.16% | ||
stahlberg | 0 | 929,164,942 | 2.16% | ||
gabrielatravels | 0 | 202,288,431 | 1.08% | ||
cordeta | 0 | 74,972,136 | 2.16% | ||
reizak | 0 | 334,100,315 | 1.72% | ||
vjap55 | 0 | 804,623,321 | 100% | ||
zlatkamrs | 0 | 224,281,101 | 4.1% | ||
monie | 0 | 422,020,950 | 100% | ||
creatrixity | 0 | 55,214,857 | 2.16% | ||
eliaschess333 | 0 | 9,715,377,306 | 50% | ||
shoganaii | 0 | 185,799,686 | 10% | ||
darkiche | 0 | 76,730,472 | 10% | ||
ydavgonzalez | 0 | 1,762,973,822 | 5% | ||
payger | 0 | 121,998,538 | 2.16% | ||
langford | 0 | 347,759,600 | 20% | ||
mattiarinaldoni | 0 | 0 | 1% | ||
mathowl | 0 | 4,783,055,581 | 10% | ||
shinedojo | 0 | 374,508,635 | 4.32% | ||
hongtao | 0 | 196,018,745 | 52% | ||
gaming.yer | 0 | 431,497,396 | 100% | ||
steem-familia | 0 | 430,091,726 | 100% | ||
lacher-prise | 0 | 179,811,678 | 10% | ||
terrylovejoy | 0 | 3,387,454,072 | 8% | ||
jcalero | 0 | 142,234,129 | 4.32% | ||
wisewoof | 0 | 118,979,641 | 2.16% | ||
vilda | 0 | 161,529,999 | 50% | ||
olajidekehinde | 0 | 81,822,455 | 10% | ||
real2josh | 0 | 153,135,951 | 10% | ||
steepup | 0 | 288,505,220 | 8% | ||
reavercois | 0 | 0 | 5% | ||
gribouille | 0 | 515,253,136 | 20% | ||
traviseric | 0 | 195,682,059 | 50% | ||
woolfe19861008 | 0 | 110,219,455 | 7.78% | ||
yrmaleza | 0 | 375,153,576 | 50% | ||
stemng | 0 | 6,617,502,941 | 10% | ||
mininthecity | 0 | 158,461,443 | 3.45% | ||
trixie | 0 | 64,900,182 | 10% | ||
kingabesh | 0 | 508,455,446 | 10% | ||
evangelista.yova | 0 | 427,613,552 | 100% | ||
miguelangel2801 | 0 | 796,864,349 | 50% | ||
dailychina | 0 | 1,984,929,892 | 7.35% | ||
didic | 0 | 2,464,267,746 | 2.16% | ||
jenniferjulieth | 0 | 391,156,556 | 100% | ||
operahoser | 0 | 227,432,926 | 0.64% | ||
emiliomoron | 0 | 3,722,017,129 | 50% | ||
dexterdev | 0 | 3,112,418,027 | 10% | ||
intellihandling | 0 | 2,944,430,412 | 50% | ||
oghie | 0 | 522,635,086 | 50% | ||
geopolis | 0 | 1,302,570,567 | 20% | ||
ajfernandez | 0 | 371,453,041 | 100% | ||
dongfengman | 0 | 725,023,072 | 7.78% | ||
robertbira | 0 | 2,099,210,928 | 5% | ||
bearded-benjamin | 0 | 58,597,344,564 | 50% | ||
alexdory | 0 | 3,986,290,395 | 8% | ||
vegan.niinja | 0 | 246,201,786 | 2.16% | ||
flugschwein | 0 | 4,656,700,386 | 19% | ||
benleemusic | 0 | 309,568,217 | 0.43% | ||
ulisesfl17 | 0 | 1,840,070,872 | 100% | ||
arac | 0 | 1,008,289,245 | 100% | ||
shentrading | 0 | 260,833,925 | 0.9% | ||
francostem | 0 | 2,751,212,193 | 20% | ||
ivan-g | 0 | 544,260,542 | 2.16% | ||
endopediatria | 0 | 695,806,150 | 20% | ||
tajstar | 0 | 65,116,267 | 100% | ||
croctopus | 0 | 1,470,110,794 | 100% | ||
zipporah | 0 | 887,800,460 | 0.86% | ||
sissyjill | 0 | 71,093,620 | 7% | ||
ingmarvin | 0 | 400,511,483 | 100% | ||
emmanuel293 | 0 | 100,883,333 | 25% | ||
cryptofuwealth | 0 | 41,362,211 | 11% | ||
ethanlee | 0 | 184,748,673 | 6.44% | ||
morbyjohn | 0 | 126,891,780 | 7% | ||
alix96 | 0 | 391,313,144 | 100% | ||
ambitiouslife | 0 | 232,710,755 | 2.16% | ||
tomastonyperez | 0 | 11,214,344,782 | 50% | ||
jingis07 | 0 | 381,963,364 | 2.16% | ||
elvigia | 0 | 9,417,363,596 | 50% | ||
scoora82 | 0 | 902,442,460 | 24% | ||
gabyoraa | 0 | 102,048,633 | 2.16% | ||
lesmouths-travel | 0 | 976,506,185 | 13% | ||
cjunros | 0 | 91,992,932 | 2.16% | ||
effofex | 0 | 2,140,127,707 | 10% | ||
luiscd8a | 0 | 1,649,097,393 | 80% | ||
lilypang22 | 0 | 160,403,594 | 7.21% | ||
samlee2018 | 0 | 59,775,358 | 50% | ||
eniolw | 0 | 369,142,144 | 5% | ||
de-stem | 0 | 7,678,735,184 | 19.8% | ||
elsll | 0 | 101,641,414 | 4.32% | ||
geadriana | 0 | 597,929,673 | 15% | ||
elpdl | 0 | 451,812,719 | 100% | ||
derbesserwisser | 0 | 1,362,803,325 | 100% | ||
serylt | 0 | 3,480,636,660 | 19.6% | ||
bavi | 0 | 127,317,250 | 2.16% | ||
hiddenblade | 0 | 1,573,198,659 | 3.45% | ||
misia1979 | 0 | 386,770,338 | 2.16% | ||
josedelacruz | 0 | 4,424,720,321 | 50% | ||
joseangelvs | 0 | 1,999,608,225 | 100% | ||
viannis | 0 | 1,609,993,251 | 50% | ||
flores39 | 0 | 398,500,814 | 100% | ||
majapesi | 0 | 251,834,827 | 50% | ||
erickyoussif | 0 | 3,532,329,158 | 100% | ||
michaelwrites | 0 | 229,567,867 | 10% | ||
deholt | 0 | 984,116,727 | 20% | ||
veteranforcrypto | 0 | 1,375,928,103 | 4.32% | ||
edanya | 0 | 77,172,446 | 2.16% | ||
goodway | 0 | 137,411,637 | 1% | ||
stevenwood | 0 | 78,709,436 | 2.16% | ||
temitayo-pelumi | 0 | 1,517,112,520 | 20% | ||
andrick | 0 | 459,960,525 | 50% | ||
sweet-jenny8 | 0 | 1,422,159,387 | 7.78% | ||
yusvelasquez | 0 | 474,571,644 | 50% | ||
doctor-cog-diss | 0 | 316,438,455 | 20% | ||
alexworld | 0 | 354,216,410 | 25% | ||
gracelbm | 0 | 200,908,258 | 2.16% | ||
frost1903 | 0 | 44,769,542 | 50% | ||
avizor | 0 | 63,580,224 | 2.16% | ||
acont | 0 | 250,171,126 | 50% | ||
niouton | 0 | 177,360,086 | 0.86% | ||
elimao | 0 | 396,117,362 | 100% | ||
schroders | 0 | 1,630,729,357 | 1.29% | ||
anaestrada12 | 0 | 17,113,760,947 | 100% | ||
robertyan | 0 | 4,462,629,242 | 100% | ||
steemzeiger | 0 | 1,082,883,866 | 19.8% | ||
yorgermadison | 0 | 354,459,383 | 100% | ||
alexjunior | 0 | 367,381,831 | 100% | ||
juliocaraballo | 0 | 65,264,661 | 50% | ||
kosheliuk | 0 | 492,259,538 | 100% | ||
ushkurwike | 0 | 492,234,293 | 100% | ||
borndead05 | 0 | 492,237,512 | 100% | ||
antunez25 | 0 | 396,041,173 | 100% | ||
haf67 | 0 | 371,659,258 | 100% | ||
joelsegovia | 0 | 3,528,022,786 | 50% | ||
chavas | 0 | 413,868,324 | 100% | ||
longer | 0 | 224,267,615 | 50% | ||
blewitt | 0 | 1,283,683,309 | 0.3% | ||
kafupraise | 0 | 83,464,520 | 34% | ||
biomimi | 0 | 191,509,014 | 40% | ||
ibk-gabriel | 0 | 121,868,651 | 10% | ||
drsensor | 0 | 1,465,891,346 | 8% | ||
mirzantorres | 0 | 250,389,145 | 50% | ||
hhtb | 0 | 166,070,303 | 2.16% | ||
jesusfl17 | 0 | 399,001,712 | 100% | ||
angelica7 | 0 | 725,738,069 | 0.21% | ||
purelyscience | 0 | 117,631,595 | 10% | ||
eglinson | 0 | 343,600,005 | 100% | ||
uzcateguiazambra | 0 | 403,958,976 | 100% | ||
yomismosoy | 0 | 234,013,924 | 50% | ||
casiloko | 0 | 216,753,485 | 50% | ||
bflanagin | 0 | 312,371,148 | 2.16% | ||
ubaldonet | 0 | 3,226,444,644 | 70% | ||
asmeira | 0 | 436,238,743 | 100% | ||
garrillo | 0 | 347,906,711 | 100% | ||
yestermorrow | 0 | 1,864,420,310 | 6% | ||
mary11 | 0 | 355,868,050 | 75% | ||
laiyuehta | 0 | 89,819,958 | 4.98% | ||
hansmast | 0 | 297,887,406 | 2.16% | ||
turtlegraphics | 0 | 652,471,825 | 7.31% | ||
wstanley226 | 0 | 75,814,990 | 50% | ||
gpcx86 | 0 | 630,659,953 | 25% | ||
amart29 | 0 | 1,706,955,098 | 20% | ||
andypalacios | 0 | 210,535,862 | 50% | ||
reinaseq | 0 | 6,168,981,877 | 100% | ||
suasteguimichel | 0 | 66,756,394 | 50% | ||
yaelg | 0 | 2,444,885,269 | 5% | ||
pfernandezpetit | 0 | 356,408,416 | 100% | ||
mgarrillogonzale | 0 | 378,710,233 | 100% | ||
rubenp | 0 | 452,354,073 | 100% | ||
jeferc | 0 | 449,191,948 | 100% | ||
clement.poiret | 0 | 237,124,482 | 4.32% | ||
lupafilotaxia | 0 | 14,836,727,747 | 100% | ||
sweatsavory | 0 | 492,251,552 | 100% | ||
americamode | 0 | 492,268,523 | 100% | ||
searchmalt | 0 | 492,237,746 | 100% | ||
fran.frey | 0 | 1,832,719,665 | 50% | ||
snoozesalsa | 0 | 492,265,781 | 100% | ||
bohrbowling | 0 | 492,247,523 | 100% | ||
doorpedal | 0 | 492,223,994 | 100% | ||
fimschell | 0 | 492,249,613 | 100% | ||
perpetuum-lynx | 0 | 415,520,844 | 19.6% | ||
alaiza | 0 | 466,520,653 | 100% | ||
jrevilla | 0 | 340,157,851 | 50% | ||
annaabi | 0 | 309,681,981 | 2.16% | ||
abraham10 | 0 | 66,152,470 | 82% | ||
alfonzoasdrubal | 0 | 550,844,335 | 100% | ||
emperorhassy | 0 | 134,980,323 | 10% | ||
moniroy | 0 | 3,331,699,357 | 50% | ||
skorup87 | 0 | 16,221,361 | 11% | ||
trang | 0 | 370,391,719 | 2.16% | ||
stem-espanol | 0 | 58,422,605,454 | 100% | ||
praditya | 0 | 869,619,254 | 24% | ||
lapp | 0 | 468,825,799 | 100% | ||
steemtpistia | 0 | 468,310,396 | 100% | ||
crassipes | 0 | 468,547,482 | 100% | ||
yashshah991 | 0 | 66,139,152 | 50% | ||
rhethypo | 0 | 106,614,492 | 2.16% | ||
predict-crypto | 0 | 94,657,805 | 0.08% | ||
chickenmeat | 0 | 195,665,223 | 2.16% | ||
javier.dejuan | 0 | 4,505,841,792 | 20% | ||
witnesstools | 0 | 627,761,450 | 7.31% | ||
sciencetech | 0 | 127,465,660 | 2% | ||
cutnadhifa | 0 | 0 | 3% | ||
faithfullwills | 0 | 67,604,213 | 85% | ||
agrovision | 0 | 468,825,407 | 100% | ||
cutnur | 0 | 0 | 1% | ||
hirally | 0 | 404,488,619 | 100% | ||
emynb | 0 | 350,210,986 | 100% | ||
fanta-steem | 0 | 1,432,157,741 | 30% | ||
giulyfarci52 | 0 | 1,061,475,939 | 50% | ||
semtroneum | 0 | 86,118,591 | 1.29% | ||
eu-id | 0 | 624,156,263 | 10% | ||
ilovecoding | 0 | 622,543,255 | 7.31% | ||
cryptowrld | 0 | 0 | 26% | ||
stem.witness | 0 | 14,405,559,741 | 20% | ||
n4dh1f4 | 0 | 0 | 1% | ||
sarhugo | 0 | 120,356,203 | 2.16% | ||
xuhi | 0 | 62,764,139 | 50% | ||
anthive | 0 | 66,900,039 | 50% | ||
andiblok | 0 | 64,801,120 | 25% | ||
steemfuckeos | 0 | 393,637,274 | 7.31% | ||
double-negative | 0 | 496,683,163 | 20% | ||
alex-hm | 0 | 949,007,152 | 50% | ||
wilmer14molina | 0 | 484,926,850 | 100% | ||
eugenialobo | 0 | 406,831,079 | 100% | ||
ballesteroj | 0 | 363,478,846 | 100% | ||
jcmontilva | 0 | 417,213,533 | 100% | ||
rodriguezr | 0 | 376,146,823 | 100% | ||
marbely20 | 0 | 402,111,073 | 100% | ||
moyam | 0 | 452,976,384 | 100% | ||
emilycg | 0 | 373,362,482 | 100% | ||
darys | 0 | 418,513,629 | 100% | ||
sibaja | 0 | 422,658,765 | 100% | ||
balcej | 0 | 452,909,015 | 100% | ||
lmanjarres | 0 | 409,859,216 | 100% | ||
anaka | 0 | 450,532,585 | 100% | ||
benhurg | 0 | 452,975,584 | 100% | ||
judisa | 0 | 451,081,551 | 100% | ||
juddarivv | 0 | 452,973,959 | 100% | ||
mariamo | 0 | 389,628,945 | 100% | ||
kimmorales | 0 | 412,712,694 | 100% | ||
loraine25 | 0 | 401,495,739 | 100% | ||
kingnosa | 0 | 51,710,600 | 50% | ||
pamahdoo | 0 | 0 | 9% | ||
kakakk | 0 | 287,097,362 | 2.16% | ||
reedhhw | 0 | 50,874,563 | 50% | ||
ascorphat | 0 | 2,297,268,964 | 2.5% | ||
cameravisual | 0 | 2,977,557,423 | 50% | ||
amin-ove | 0 | 66,042,327 | 50% | ||
huilco | 0 | 278,267,322 | 100% | ||
hanyseek | 0 | 3,156,607 | 25% | ||
cerd26 | 0 | 85,219,243 | 100% | ||
herculean | 0 | 66,823,145 | 50% | ||
combatsports | 0 | 1,331,530,612 | 4.32% | ||
jent | 0 | 1,715,567,679 | 70% | ||
donasys | 0 | 47,003,169 | 50% | ||
mtfmohammad | 0 | 100,917,646 | 25% | ||
cleiver | 0 | 51,533,402 | 50% | ||
thewhalehunter | 0 | 53,831,355 | 50% | ||
chrisluke | 0 | 107,132,775 | 26% | ||
joannar | 0 | 103,570,640 | 25% | ||
pflanzenlilly | 0 | 221,900,095 | 50% | ||
laithut | 0 | 451,441,357 | 100% | ||
duromyed | 0 | 451,505,614 | 100% | ||
fararlue | 0 | 451,467,922 | 100% | ||
oryow | 0 | 451,395,158 | 100% | ||
dooworisa | 0 | 451,380,948 | 100% | ||
smalltall | 0 | 704,385,886 | 20% | ||
faberleggenda | 0 | 295,874,311 | 100% | ||
mohaaking | 0 | 71,496,002 | 50% | ||
celine-robichaud | 0 | 103,003,686 | 50% | ||
imaloser | 0 | 51,721,816 | 50% |
Congratulations @hongtao! Your post was mentioned in the [Steem Hit Parade for newcomers](https://steemit.com/hit-parade/@arcange/daily-hit-parade-for-newcomers-20190319) in the following category: * Upvotes - Ranked 9 with 385 upvotes I also upvoted your post to increase its reward If you like my work to promote newcomers and give them more visibility on the Steem blockchain, consider to [vote for my witness](https://steemit.com/~witnesses)!
author | arcange |
---|---|
permlink | re-tensorflow-overfitting-20190319t170645000z |
category | cn-stem |
json_metadata | "" |
created | 2019-03-20 16:07:30 |
last_update | 2019-03-20 16:07:30 |
depth | 1 |
children | 2 |
last_payout | 2019-03-27 16:07:30 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 448 |
author_reputation | 1,146,606,601,469,178 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,629,315 |
net_rshares | 0 |
author | hongtao |
---|---|
permlink | re-arcange-re-tensorflow-overfitting-20190320t161214776z |
category | cn-stem |
json_metadata | {"tags":["cn-stem"],"app":"steemit/0.1"} |
created | 2019-03-20 16:12:15 |
last_update | 2019-03-20 16:12:15 |
depth | 2 |
children | 1 |
last_payout | 2019-03-27 16:12:15 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.050 HBD |
curator_payout_value | 0.016 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 64 |
author_reputation | 3,241,267,862,629 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,629,547 |
net_rshares | 101,342,221,916 |
author_curate_reward | "" |
voter | weight | wgt% | rshares | pct | time |
---|---|---|---|---|---|
arcange | 0 | 96,792,345,719 | 10% | ||
raphaelle | 0 | 4,549,876,197 | 10% |
Thank you for you support and witness vote @Hongtao, really appreciated.
author | arcange |
---|---|
permlink | re-hongtao-re-arcange-re-tensorflow-overfitting-20190320t170043839z |
category | cn-stem |
json_metadata | {"tags":["cn-stem"],"users":["hongtao"],"app":"steemit/0.1"} |
created | 2019-03-20 17:00:45 |
last_update | 2019-03-20 17:00:45 |
depth | 3 |
children | 0 |
last_payout | 2019-03-27 17:00:45 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 72 |
author_reputation | 1,146,606,601,469,178 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,631,346 |
net_rshares | 0 |
帅哥/美女!想要参加活动但是不知道从何开始?关注寻宝团@cn-activity每日整理社区活动!倘若你想让我隐形,请回复“取消”。
author | cnbuddy |
---|---|
permlink | re-hongtao-tensorflow-overfitting-20190319t174009158z |
category | cn-stem |
json_metadata | "" |
created | 2019-03-19 17:40:09 |
last_update | 2019-03-19 17:40:09 |
depth | 1 |
children | 0 |
last_payout | 2019-03-26 17:40:09 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 65 |
author_reputation | -1,449,160,991,441 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,582,432 |
net_rshares | 0 |
提一些建议哦: 1. Github repo似乎重命名了文件夹,所以无法访问了,可以更新一下链接; 2. 可以最后加一个参考/引用的部分,包括:使用了哪些使用的公开数据(如Kaggle Dataset, US Census), 参考代码(如果有的话),参考文档或书籍(如TensorFlow 或 Keras的文档,如果值得读者去了解的话); 3. 不太清楚Steem的markdown是否有可能显示Jupyter Notebook或者RMarkdown,可以直接嵌入的话可能效果更好。(可能需要作者做些调研) 4. 内容方面,我觉得可以加入更多自己的实验或者批判,如果只是把最基本的回归、分类、过拟合等问题介绍一遍,我觉得效果可能不如直接看文档或教材。所以,最好能添加更多自己的想法,比如学习的感悟和建议等等,这对于作者的学习以及读者的借鉴,都会有所助益。
author | robertyan |
---|---|
permlink | re-hongtao-tensorflow-overfitting-20190320t072747208z |
category | cn-stem |
json_metadata | {"tags":["cn-stem"],"app":"steemit/0.1"} |
created | 2019-03-20 07:27:48 |
last_update | 2019-03-20 07:46:09 |
depth | 1 |
children | 2 |
last_payout | 2019-03-27 07:27:48 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 379 |
author_reputation | 16,909,391,530,163 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,610,830 |
net_rshares | 455,229,484 |
author_curate_reward | "" |
voter | weight | wgt% | rshares | pct | time |
---|---|---|---|---|---|
hongtao | 0 | 455,229,484 | 100% |
谢谢建议, 1. 昨天重新commit 了代码,整理了子目录,忘了改链接地址了,已经将子目录替换成了项目地址。 2.公开数据的来源在源代码中已经做了说明,接受建议,在文末添加了。 3. Steemit的Markdown确实对代码支持不太友好,没有语法高亮,目前除了截图好像还没找到其他都方法。 4. 本人在这个领域也是初学者,我会在后续的文章中尽量加入自己的思考
author | hongtao |
---|---|
permlink | re-robertyan-re-hongtao-tensorflow-overfitting-20190320t102155337z |
category | cn-stem |
json_metadata | {"tags":["cn-stem"],"app":"steemit/0.1"} |
created | 2019-03-20 10:21:54 |
last_update | 2019-03-20 10:21:54 |
depth | 2 |
children | 1 |
last_payout | 2019-03-27 10:21:54 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 181 |
author_reputation | 3,241,267,862,629 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,616,129 |
net_rshares | 4,418,924,261 |
author_curate_reward | "" |
voter | weight | wgt% | rshares | pct | time |
---|---|---|---|---|---|
robertyan | 0 | 4,418,924,261 | 100% |
哈哈,不客气,我只是从写此类分享文章的角度提一下自己的看法。😀 Steemit代码展示我也有些不了解,晚些看到方法我分享一下😛 你有加入新手村吗?可以联系一下村长 [@ericet](https://steemit.com/@ericet) 他的微信也是ericet
author | robertyan |
---|---|
permlink | robertyan-re-hongtao-re-robertyan-re-hongtao-tensorflow-overfitting-20190320t104058198z |
category | cn-stem |
json_metadata | {"app":"partiko","client":"ios"} |
created | 2019-03-20 10:41:45 |
last_update | 2019-03-20 10:43:36 |
depth | 3 |
children | 0 |
last_payout | 2019-03-27 10:41:45 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 135 |
author_reputation | 16,909,391,530,163 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,616,711 |
net_rshares | 0 |
结尾漏了同步到你的简书。。。
author | softmetal |
---|---|
permlink | re-hongtao-tensorflow-overfitting-20190319t174728357z |
category | cn-stem |
json_metadata | {"tags":["cn-stem"],"app":"steemit/0.1"} |
created | 2019-03-19 17:47:27 |
last_update | 2019-03-19 17:47:27 |
depth | 1 |
children | 1 |
last_payout | 2019-03-26 17:47:27 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 14 |
author_reputation | 146,291,910,015,030 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,582,702 |
net_rshares | 0 |
哈哈 谢谢提醒 Posted using [Partiko iOS](https://partiko.app/referral/hongtao)
author | hongtao |
---|---|
permlink | hongtao-re-softmetal-re-hongtao-tensorflow-overfitting-20190319t190356605z |
category | cn-stem |
json_metadata | {"app":"partiko","client":"ios"} |
created | 2019-03-19 19:04:03 |
last_update | 2019-03-19 19:04:03 |
depth | 2 |
children | 0 |
last_payout | 2019-03-26 19:04:03 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.000 HBD |
curator_payout_value | 0.000 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 73 |
author_reputation | 3,241,267,862,629 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,585,303 |
net_rshares | 0 |
<div class='text-justify'> <div class='pull-left'> <br /> <center> <img width='125' src='https://i.postimg.cc/9FwhnG3w/steemstem_curie.png'> </center> <br/> </div> <br /> <br /> This post has been voted on by the **SteemSTEM** curation team and voting trail in collaboration with **@curie**. <br /> If you appreciate the work we are doing then consider [voting](https://www.steemit.com/~witnesses) both projects for witness by selecting [**stem.witness**](https://steemconnect.com/sign/account_witness_vote?approve=1&witness=stem.witness) and [**curie**](https://steemconnect.com/sign/account_witness_vote?approve=1&witness=curie)! <br /> For additional information please join us on the [**SteemSTEM discord**]( https://discord.gg/BPARaqn) and to get to know the rest of the community! </div>
author | steemstem |
---|---|
permlink | re-hongtao-tensorflow-overfitting-20190320t131706039z |
category | cn-stem |
json_metadata | {"app":"bloguable-bot"} |
created | 2019-03-20 13:17:12 |
last_update | 2019-03-20 13:17:12 |
depth | 1 |
children | 0 |
last_payout | 2019-03-27 13:17:12 |
cashout_time | 1969-12-31 23:59:59 |
total_payout_value | 0.593 HBD |
curator_payout_value | 0.197 HBD |
pending_payout_value | 0.000 HBD |
promoted | 0.000 HBD |
body_length | 800 |
author_reputation | 262,017,435,115,313 |
root_title | Tensorflow入门——处理overfitting的问题 |
beneficiaries | [] |
max_accepted_payout | 1,000,000.000 HBD |
percent_hbd | 10,000 |
post_id | 81,622,131 |
net_rshares | 1,178,205,220,694 |
author_curate_reward | "" |
voter | weight | wgt% | rshares | pct | time |
---|---|---|---|---|---|
delegate.lafona | 0 | 373,830,430,233 | 20% | ||
steemstem | 0 | 788,694,668,431 | 20% | ||
stem.witness | 0 | 13,668,068,236 | 20% | ||
ascorphat | 0 | 2,012,053,794 | 2.5% |