CGNN复现遇到的问题及解决办法（持续更新中）

论文：Causal Generative Neural Networks

1、运行报错

Q1

• Q: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library ‘cudart64_110.dll’; dlerror: cudart64_110.dll not found
  I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
• A: 下载cudart64_110.dll，放到项目路径下

Q2

• Q: Traceback (most recent call last):File “D:/PycharmWorkSpace/CGNN-master/run_CGNN_graph.py”, line 1, in
  import cgnn
  File “d:\pycharmworkspace\cgnn-master\code\cgnn_init_.py”, line 6, in
  from .generators import______init______
  File “d:\pycharmworkspace\cgnn-master\code\cgnn\generators_init_.py”, line 1, in
  from .random_graph_generator import RandomGraphGenerator
  File “d:\pycharmworkspace\cgnn-master\code\cgnn\generators\random_graph_generator.py”, line 1, in
  from .functions_default import (noise, cause, effect, rand_bin)
  File “d:\pycharmworkspace\cgnn-master\code\cgnn\generators\functions_default.py”, line 14, in
  from sklearn.mixture import GMM
  ImportError: cannot import name ‘GMM’

• A: 查阅对应 scikit-learn 版本的官方文档后发现，GMM 被替换为 GaussianMixture，故正确写法为：

  from sklearn.mixture import GaussianMixture
  

  启示：

  遇到冷门报错信息，束手无策时，可以考虑查查官方文档

  转载于:https://www.cnblogs.com/xiangsui/p/10981529.html

Q3

• Q: Processing Example_graph_numdata.csv…Traceback (most recent call last):File “D:/PycharmWorkSpace/CGNN-master/run_CGNN_graph.py”, line 22, in
  p_directed_graph = GNN.orient_graph(data, umg, printout=datafile + ‘_printout.csv’)
  File “d:\pycharmworkspace\cgnn-master\code\cgnn\PairwiseModel.py”, line 75, in orient_graph
  weight = self.predict_proba(scale(df_data[a].as_matrix()), scale(df_data[b].as_matrix()),idx)
  File “D:\Software\Anaconda\envs\cgnn\lib\site-packages\pandas\core\generic.py”, line 5141, in getattr
  return object.getattribute(self, name)
  AttributeError: ‘Series’ object has no attribute ‘as_matrix’

• A: 出错原因：as_matrix不能用

  解决方式：将.as_matrix()改为.values

  测试运行：

Q4

• Q: AttributeError: module ‘tensorflow’ has no attribute ‘placeholder’
• A: tensorflow版本不对，在2.+的版本下使用了1.+的接口，修改tensorflow的版本为1.15.0