1: \begin{abstract}

2: With several advantages and as an alternative to predict physics field, 

3: machine learning methods can be classified into two distinct types: data-driven relying on training data and physics-driven using physics law. 

4: Choosing heat conduction problem as an example, we compared the data- and physics-driven learning process with deep Convolutional Neural Networks (CNN). 

5: It shows that the convergences of the error to ground truth solution and the residual of heat conduction equation exhibit remarkable differences. 

6: Based on this observation, we propose a combined-driven method for learning acceleration and more accurate solutions.

7: With a weighted loss function, reference data and physical equation are able to simultaneously drive the learning. 

8: Several numerical experiments are conducted to investigate the effectiveness of the combined method. 

9: For the data-driven based method, the introduction of physical equation not only is able to speed up the convergence, but also produces physically more consistent solutions. 

10: For the physics-driven based method, it is observed that the combined method is able to speed up the convergence up to 49.0\% by using a not very restrictive coarse reference.

11: \end{abstract}

12: