- numpy v1.12
- sklearn
The MNIST database of handwritten digits
Reference : http://yann.lecun.com/exdb/mnist/
usage: main.py [-h] [--train_X TRAIN_X] [--train_T TRAIN_T] [--test_X TEST_X]
[--test_T TEST_T] [--load LOAD] [--save SAVE] [--pre {pca,lda}]
[--deg DEG] [--task {validate,train,plot}] [--model {rf}]
[--param_rf PARAM_RF] [--verbose VERBOSE]
optional arguments:
-h, --help show this help message and exit
--train_X TRAIN_X training data X
--train_T TRAIN_T training data T
--test_X TEST_X testing data X
--test_T TEST_T testing data T
--load LOAD model load from
--save SAVE model save to
--pre {pca,lda} preprocess type
--deg DEG degree for preprocess
--task {validate,train,plot}
task type
--model {rf} model type
--param_rf PARAM_RF parameter for rf
--verbose VERBOSE log on/off
./train.sh "{list of number of decision tree}" "{list of minimum number of samples per leaf nodes}" "list of fraction of samples"
e.g.
./train.sh "100 1000" "10 5" "0.25 0.5"