Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledge

We consider the use of Deep Learning methods for modeling complex phenomena those occurring in natural physical processes. With the large amount of gathered on these phenomena the data intensive paradigm could begin to more traditional approaches elaborated over the years in fields like or physics. However, despite considerable successes in a variety of domains, the machine learning field is not yet ready to handle the of complexity required by such problems. Using an example application, Sea Surface Temperature Prediction, we show how general background gained from physics could be used as a guideline for designing Deep Learning models. In order to motivate the approach and to assess generality we demonstrate a formal link between the solution of a class of equations underlying a large family of physical phenomena and the model. Experiments and comparison with series of baselines including a of the art numerical approach is then provided.