deep learning - h2o R samples

用H2O做了个deep learning的例子（0、1二分类），对比logistic回归、GAM、SVM等，综合表现最差，比较郁闷。deep learning中可以调节的参数很多，不确定是参数问题还是其它问题。有了解的童鞋吗？

R程序：

(1) library(h2o) localH2O = h2o.init() woedata.h2o <- as.h2o(woedata) woemdl.dl <- h2o.deeplearning(x = 2:5, y = 1, training_frame = woedata.h2o, score_training_samples = 0) woemdl.dl prd.dl <- h2o.predict(woemdl.dl, woedata.h2o) h2o.table(prd.dl[,1], woedata.h2o[,1])

(2) trainData.h2o <- as.h2o(trainData) trainmdl.dl <- h2o.deeplearning(x = 1:4, y = 5, training_frame = trainData.h2o, score_training_samples = 0) trainmdl.dl

(3) woemdl.dl <- h2o.deeplearning(x = 2:5, y = 1, training_frame = woedata.h2o, score_training_samples = 0,activation="Maxout") woemdl.dl

(4) woemdl.dl <- h2o.deeplearning(x = 2:5, y = 1, training_frame = woedata.h2o, score_training_samples = 0,activation="MaxoutWithDropout") woemdl.dl

(5) woemdl.dl <- h2o.deeplearning(x = 2:5, y = 1, training_frame = woedata.h2o, score_training_samples = 0,activation="Maxout", hidden = c(350, 200), epochs=10) woemdl.dl

(6) woemdl.dl <- h2o.deeplearning(x = 2:5, y = 1, training_frame = woedata.h2o, score_training_samples = 0,activation="RectifierWithDropout", input_dropout_ratio=0, hidden_dropout_ratios=c(0.2,0.1,0.1,0), l1=1e-5, l2=1e-5, hidden = c(200,200,200,200), epochs=15)

 

结果：

(1)           0     1    Error         Rate 0      5790  7941 0.578326  =7941/13731 1      1446  4473 0.244298   =1446/5919 Totals 7236 12414 0.477710  =9387/19650     predict y count 1       1 1  4473 2       0 0  5790 3       0 1  1446 4       1 0  7941

(2)           0     1    Error         Rate 0      5787  7944 0.578545  =7944/13731 1      1357  4562 0.229262   =1357/5919 Totals 7144 12506 0.473333  =9301/19650

(3)           0     1    Error         Rate 0      7414  6317 0.460054  =6317/13731 1      2060  3859 0.348032   =2060/5919 Totals 9474 10176 0.426310  =8377/19650

(4)           0     1    Error         Rate 0      6875  6856 0.499308  =6856/13731 1      1831  4088 0.309343   =1831/5919 Totals 8706 10944 0.442087  =8687/19650

(5)           0     1    Error         Rate 0      7501  6230 0.453718  =6230/13731 1      2037  3882 0.344146   =2037/5919 Totals 9538 10112 0.420712  =8267/19650

(6)           0     1    Error         Rate 0      6669  7062 0.514311  =7062/13731 1      1779  4140 0.300558   =1779/5919 Totals 8448 11202 0.449924  =8841/19650