boyangumn / dcn Goto Github PK

I've executed the run_raw_mnist.py with 'hidden_dim': [500, 500, 2000, 10] like described in the paper.

THEANO_FLAGS='floatX=float32' python run_raw_mnist.py

But I got lower NMI, ARI, ACC than expected:

NMI for deep clustering: 0.48
ARI for deep clustering: 0.35
ACC for deep clustering: 0.52

Is this the expected result? The results in the paper are, respectively, around: 0.81, 0.75 and 0.83

I am trying to run this code. But this error happens:

python2 run_20News.py
WARNING (theano.tensor.blas): Using NumPy C-API based implementation for BLAS functions.
Traceback (most recent call last):
File "run_20News.py", line 17, in
from multi_layer_km import test_SdC, load_data
File "/home/code/kmean-deep/DCN/multi_layer_km.py", line 24, in
from dA import dA
ImportError: No module named dA

Hi,
I have tried running your run_raw_mnist.py and run_pendigits.py and I am obtaining the same error for both the experiments. The error is show

I am not well versed with theano. So if you could let me know if I am doing something wrong or is there is any changes in the setting that I have to look into

Thanks

Hi

I was going through your paper and decided to run your code. But I am getting this error. I guess u haven't provided the dA.py file. I have copied the commented code from dA_init.py to remedy nthe

Thanks
Soumava Kumar Roy