magnumical / gcn_for_eeg Goto Github PK

Dear Scholar,
Foremost, I will like to appreciate your intellectual effort in putting up the GCN related codes as compact as possible, it has really helped in conceiving the idea greatly.

I tried to execute the onEEGcode but using the DenseGCN_Model instead of the the models used by default and surprisingly, I got the error below

Traceback (most recent call last):
File "C:/Users/Abdul/PycharmProjects/AbdulProject2/03Main/main_onEEGcode.py", line 89, in
model=DenseGCN_Model.cgcnn(L, **params)
File "C:\Users\Abdul\PycharmProjects\AbdulProject2\scripts\DenseGCN_Model.py", line 435, in init
self.build_graph(M_0)
File "C:\Users\Abdul\PycharmProjects\AbdulProject2\scripts\DenseGCN_Model.py", line 165, in build_graph
op_logits = self.inference(self.ph_data, self.ph_dropout)
File "C:\Users\Abdul\PycharmProjects\AbdulProject2\scripts\DenseGCN_Model.py", line 193, in inference
logits = self._inference(data, dropout)
File "C:\Users\Abdul\PycharmProjects\AbdulProject2\scripts\DenseGCN_Model.py", line 703, in _inference
x_2 = self.filter(x_1, self.L[1], self.F[1], self.K[1])
File "C:\Users\Abdul\PycharmProjects\AbdulProject2\scripts\DenseGCN_Model.py", line 546, in chebyshev5
x1 = tf.sparse_tensor_dense_matmul(L, x0)
File "C:\Users\Abdul\Anaconda2\envs\AbdulProject2\lib\site-packages\tensorflow\python\ops\sparse_ops.py", line 2336, in sparse_tensor_dense_matmul
adjoint_b=adjoint_b)
File "C:\Users\Abdul\Anaconda2\envs\AbdulProject2\lib\site-packages\tensorflow\python\ops\gen_sparse_ops.py", line 2910, in sparse_tensor_dense_mat_mul
adjoint_b=adjoint_b, name=name)
File "C:\Users\Abdul\Anaconda2\envs\AbdulProject2\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 788, in _apply_op_helper
op_def=op_def)
File "C:\Users\Abdul\Anaconda2\envs\AbdulProject2\lib\site-packages\tensorflow\python\util\deprecation.py", line 507, in new_func
return func(*args, **kwargs)
File "C:\Users\Abdul\Anaconda2\envs\AbdulProject2\lib\site-packages\tensorflow\python\framework\ops.py", line 3300, in create_op
op_def=op_def)
File "C:\Users\Abdul\Anaconda2\envs\AbdulProject2\lib\site-packages\tensorflow\python\framework\ops.py", line 1823, in init
control_input_ops)
File "C:\Users\Abdul\Anaconda2\envs\AbdulProject2\lib\site-packages\tensorflow\python\framework\ops.py", line 1662, in _create_c_op
raise ValueError(str(e))
ValueError: Dimensions must be equal, but are 32 and 64 for 'conv2/SparseTensorDenseMatMul/SparseTensorDenseMatMul' (op: 'SparseTensorDenseMatMul') with input shapes: [992,2], [992], [2], [64,16384] and with input tensors computed as partial shapes: input[2] = [32,32].

Could you please let me know the line of the DenseGCN_Model to alter to be able to solve this issue.

Thanks and regards

Sorry to bother you! I'd like to ask something about dataset converting.
I am conducting research on EEG, and I read your github code, which is very enlightening, but it also caused some problems.

1. If only a small amount but not all of the data you provide is used, do I only need to change the size of the matrix? It would be great if you could give an example.
2. How should we cite other data? Including different data formats, different number of EEG channels and different matrix sizes?
   Thank you for reading and answering, thank you.

Thank you for your previous reply, you are really a respectable scholar.
I don't know if it is a problem with my computer or the processing steps. After the .csv file is generated and the onEEGcode.py program is run, an error of “ValueError: could not convert string to float” will be reported. But after I checked, I didn't find any structure other than string. How should I modify the program?
The place where the error occurred is at “onEEGcode.py”->"train_data, train_labels, test_data, test_labels= dataread(DIR)"->"dataread.py"->"train_data = np.array(train_data).astype('float32')"
thank you for your reply.

Hello, researcher
I got your mat file in the last Issues. Then I ran onEEGcode.py, but the accuracy of the validation set was only 24.8, and after 100 epochs, the accuracy and los remained unchanged. Is this a problem with your model?

Hello friend,
I have got this problem.
(Traceback (most recent call last):
File "edfread.py", line 225, in
X, Y = load_raw_data(electrodes=electrodes, subject=subject, num_classes=nclasses)
File "edfread.py", line 169, in load_raw_data
return np.array(trials, dtype=np.float64).reshape((len(trials),) + trials[0].shape + (1,)), np.array(labels, dtype=np.float64)
IndexError: list index out of range.)
Can you give me some advice Thanks anyway!

Is there any way to optimize/fast the execution time of main code? It is taking a hell lot of time for a single epoch.
Running on a 16GB machine.

Hello, sorry to disturb you, when I saw how to read the data in your GCN_for_EEG, it is all the fourth category, the training accuracy is 0%, that is, in the first edfread.py, the labels seem to be fully assigned It is [0,0,0,1], I hope you can answer my question. Is it my problem? Is your result normal? Thank you

I have download the efd documents by using downloaddata.py. However, when I use the edfread.py to excate the edf, an error has occurred.
return np.array(trials, dtype=np.float64).reshape((len(trials),) + trials[0].shape + (1,)), \

IndexError: list index out of range

when I debug it, I found that the edf can't be opened.

Do you know how to debug it? Or would you mind sharing the complied .mat documents. Thank you for your helping and apologies for disturbing you!

scholar
Hello!
Could you please tell me, have you ever encountered a situation where the validation accuracy has always been the same value during operation? Is this my data processing problem, or some other situation, if you can give me an answer, I would be very grateful.