cyberzhg/keras-gcn

Graph convolutional layers

keras-gcn's Issues

Error when passing the arguments

Describe the Bug

I am getting the following error 👍 Traceback (most recent call last):

  File "/home/pulkit/Desktop/GIT2/BTP/SimGNN/src/model.py", line 97, in <module>
    model.train_step()
  File "/home/pulkit/Desktop/GIT2/BTP/SimGNN/src/model.py", line 88, in train_step
    y = self(data, training = True)
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/keras/engine/base_layer.py", line 985, in __call__
    outputs = call_fn(inputs, *args, **kwargs)
  File "/home/pulkit/Desktop/GIT2/BTP/SimGNN/src/model.py", line 46, in call
    z1 = self.gcn1([features_1, edge_index_1])
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/keras/engine/base_layer.py", line 985, in __call__
    outputs = call_fn(inputs, *args, **kwargs)
  File "/home/pulkit/.local/lib/python3.8/site-packages/keras_gcn/layers.py", line 50, in call
    outputs = self.activation(self._call(features, edges))
  File "/home/pulkit/.local/lib/python3.8/site-packages/keras_gcn/layers.py", line 145, in _call
    return K.batch_dot(K.permute_dimensions(edges, (0, 2, 1)), features) \
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/util/dispatch.py", line 201, in wrapper
    return target(*args, **kwargs)
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/keras/backend.py", line 2943, in permute_dimensions
    return array_ops.transpose(x, perm=pattern)
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/util/dispatch.py", line 201, in wrapper
    return target(*args, **kwargs)
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/ops/array_ops.py", line 2188, in transpose
    return transpose_fn(a, perm, name=name)
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/ops/gen_array_ops.py", line 11529, in transpose
    return transpose_eager_fallback(
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/ops/gen_array_ops.py", line 11554, in transpose_eager_fallback
    _result = _execute.execute(b"Transpose", 1, inputs=_inputs_flat,
  File "/home/pulkit/.local/lib/python3.8/site-packages/tensorflow/python/eager/execute.py", line 59, in quick_execute
    tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
tensorflow.python.framework.errors_impl.InvalidArgumentError: transpose expects a vector of size 2. But input(1) is a vector of size 3 [Op:Transpose]

The input sizes are :
Features : (19, 16)
Edges : (2, 158)

Calling the layer as : z1 = self.gcn1([features_1, edge_index_1])

代码实现的paper

@CyberZHG 感谢实现相关代码！请问你能提供关于这个代码实现的参考博客或者论文吗？

Graph convolution formula is inconsistent with the kipf paper

Hello，which formula does the formula in your code come from, which is different from the formula in the kipf paper?

In readme.md

Is the
from keras_gru import GraphConv

means
from keras_gcn import GraphConv??
is this a mistyping?

Paper reference or more explanation?

Hi,
Apparently, your codes look very efficient.
But I wonder which paper is your code based on, especially because the only function you provided is a little bit different from the original GCN (http://tkipf.github.io/graph-convolutional-networks/).

Cheers

I have the following error

---

Layer (type) Output Shape Param # Connected to

input_1 (InputLayer) [(None, None, 256)] 0 []

input_2 (InputLayer) [(None, 256, 256)] 0 []

graph_conv (GraphConv) (None, 256, 32) 8224 ['input_1[0][0]',
'input_2[0][0]']

graph_conv_1 (GraphConv) (None, 256, 16) 528 ['graph_conv[0][0]',
'input_2[0][0]']

lstm (LSTM) (None, None, 128) 197120 ['input_1[0][0]']

flatten (Flatten) (None, 4096) 0 ['graph_conv_1[0][0]']

lstm_1 (LSTM) (None, 32) 20608 ['lstm[0][0]']

concatenate (Concatenate) (None, 4128) 0 ['flatten[0][0]',
'lstm_1[0][0]']

batch_normalization (BatchNorm (None, 4128) 16512 ['concatenate[0][0]']
alization)

dropout (Dropout) (None, 4128) 0 ['batch_normalization[0][0]']

dense (Dense) (None, 128) 528512 ['dropout[0][0]']

dense_1 (Dense) (None, 32) 4128 ['dense[0][0]']

dropout_1 (Dropout) (None, 32) 0 ['dense_1[0][0]']

dense_2 (Dense) (None, 1) 33 ['dropout_1[0][0]']

==================================================================================================
Total params: 775,665
Trainable params: 767,409
Non-trainable params: 8,256

---

========== TRAIN ============
/usr/local/lib/python3.7/dist-packages/nilmtk/elecmeter.py:432: UserWarning: The provided sample_period (1) is shorter than the meter's sample_period (3)
sample_period, default_sample_period
Epoch 1/10
Traceback (most recent call last):
File "/content/drive/MyDrive/Auto_Encoder/DAE/redd-test.py", line 31, in
dae.train(train_mains, train_meter, epochs=10, sample_period=1)
File "/content/drive/MyDrive/Auto_Encoder/DAE/daedisaggregator.py", line 78, in train
self.train_on_chunk(mainchunk, meterchunk, epochs, batch_size)
File "/content/drive/MyDrive/Auto_Encoder/DAE/daedisaggregator.py", line 115, in train_on_chunk
self.model.fit(X_batch, Y_batch, batch_size=batch_size, epochs=epochs, shuffle=True)
File "/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py", line 67, in error_handler
raise e.with_traceback(filtered_tb) from None
File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py", line 1147, in autograph_handler
raise e.ag_error_metadata.to_exception(e)
ValueError: in user code:

File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 1021, in train_function  *
    return step_function(self, iterator)
File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 1010, in step_function  **
    outputs = model.distribute_strategy.run(run_step, args=(data,))
File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 1000, in run_step  **
    outputs = model.train_step(data)
File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 859, in train_step
    y_pred = self(x, training=True)
File "/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py", line 67, in error_handler
    raise e.with_traceback(filtered_tb) from None
File "/usr/local/lib/python3.7/dist-packages/keras/engine/input_spec.py", line 200, in assert_input_compatibility
    raise ValueError(f'Layer "{layer_name}" expects {len(input_spec)} input(s),'

ValueError: Layer "model" expects 2 input(s), but it received 1 input tensors. Inputs received: [<tf.Tensor 'IteratorGetNext:0' shape=(None, 256, 1) dtype=float32>]

Explanation

Hello!

I am very interested in graph convolution, especially when applied to chemical compounds.

Is there a paper, or some sort of intuitive explanation you can give of how the graph convolution layer works?

Thanks for your time

How to fit the model?

I see that there is a test module. Could you please also include an example on how to fit the model?