I have just read ten papers on deep neural networks that G. Hinton at Toronto and his group published since 2006. Insights of the deep NNs model let me hint out a direction by which we can apply what has been accumulated in various other fields of study in the area of computational intelligence. Some people among the area are very much negative on the future of computational intelligence rather than deep NNs modeling. I find, however, that is too early a decision to think about it. I do move into using deep NNs model from now on; they provide you with a very sharp knife to reduce a rich dimension of representations without or with a small amount of labeled data. I would bring what I've studied on heuristics that is to be included into computational intelligence. How would I do? I'll give a talk at conferences that are coming in November and December.