In this chapter, we'll look at some emerging Neural Network (NN) designs. They haven't reached maturity yet, but hold potential for the future because they try to address fundamental limitations in existing DL algorithms. If one day any of these technologies prove successful and useful for practical applications, we might get one step closer to artificial general intelligence.
One thing that we need to bear in mind is the nature of structured data. So far in this book, we've focused on processing either images or text—in other words, unstructured data. This is not a coincidence, because NNs excel in the seemingly complex task of finding structure in combinations of pixels or text sequences. On the other hand, ML algorithms, such as gradient boosted trees or random forests, seem to perform on a par with, or better than, NNs...
