When comparing the three comprehensive machine learning libraries highlighted in this chapter (Keras, H2O, and MXNet), there are three primary differences: external language dependencies, functions, and syntax (ease of use and cognitive load). We will now cover each of these main differences in turn.
The first major difference between the three packages is the external language dependencies for each. As mentioned earlier, none of these packages are written in R. What this means is that you will need additional languages installed on your machine in order for these packages to work. It also means that you cannot easily look at the source documentation to see how a particular function works or why you are receiving a certain error (unless you know one of the languages, of course). The packages are written using the following languages: Keras...
