参考：Keras - Python Deep Learning Neural Network API

1. Keras With TensorFlow Prerequisites - Getting Started With Neural Networks
2. TensorFlow And Keras GPU Support - CUDA GPU Setup
3. Keras With TensorFlow - Data Processing For Neural Network Training
4. Create An Artificial Neural Network With TensorFlow's Keras API
5. Train An Artificial Neural Network With TensorFlow's Keras API
6. Build A Validation Set With TensorFlow's Keras API
7. Neural Network Predictions With TensorFlow's Keras API
8. Create A Confusion Matrix For Neural Network Predictions
9. Save And Load A Model With TensorFlow's Keras API

numpy.reshape(-1, 2)

个人理解：如果出现负数，则说明这个数字可以不参考，只参考正数的部分计算即可。

a = np.array([2, 3, 4, 6])

a.reshape(-1,1)
output:
array([[2],
       [3],
       [4],
       [6]])

a.reshape(-1,2)
ouput:
array([[2, 3],
       [4, 6]])