DCGAN generating numbers

Question

Is it possible to generate values using DCGAN or it's only specified for images? here is my dataset example,

 time   lat          lon     office h-room1 h-room2 kitchen 
 6:00   48.335827   14.32085    0   1   0   0   
 7:00   48.335827   14.32085    0   0   0   1   
 8:00   48.335827   14.32085    0   0   1   0   
 9:00   48.368396   14.514999   1   0   0   0   
 10:00  48.368396   14.514999   1   0   0   0   
 11:00  48.368396   14.514999   1   0   0   0   
 12:00  48.368396   14.514999   1   0   0   0   
 13:00  48.368396   14.514999   1   0   0   0   
 14:00  48.368396   14.514999   1   0   0   0   
 15:00  48.368396   14.514999   1   0   0   0   
 16:00  48.368396   14.514999   1   0   0   0

I am getting error ,
Negative dimension size caused by subtracting 3 from 1 for 'conv2d_2/convolution while trying DCGAN for these values.

Or I should use Variational Autoencoder for values generation ?

score 0 · Answer 1 · answered Mar 21 '18 at 20:19

0

It's a Time series data so we can generate it using Recurrent Conditional GANS. check GitHub for more understanding and research papers.

answered Mar 21 '18 at 20:19

Shahek

83
1
13

DCGAN generating numbers

1 Answers1