Machine Learning for Civil and Environmental Engineers

11 Final Thoughts and Future Directions

Just because something works does not mean that it cannot be improved.

Shuri (Black Panther)

Synopsis

This chapter concludes the body of our textbook with a look into some of the future directions and research needs. Some of the discussed ideas stem directly from the limitations and challenges we have seen over the previous chapters, while others build upon those noted as identified in the open literature.

11.1 Now

Let us look at some of the key points we learned throughout this book. There are plenty of ML algorithms, and these primarily fall under three main types of learning (supervised, unsupervised, and semi-supervised). The collection of algorithms, as well as learning strategies, can find a home in many problems and areas that belong to the civil and environmental engineering discipline. The above has been common knowledge for quite some time (+20 years) [1, 2]. The adoption of machine learning (ML) has just started to prosper.

This could be due to the fact that ML is fundamentally different from the other methods we are familiar with. For example, to use or apply ML, our engineers have to learn the principles of ML. The majority of such principles remain absent from our education, and some revolve around coding and programming – an exercise that we rarely cover in our curriculum. This is further darkened by the opaque nature of most blackbox ML methods.

As you can see, we are already standing at a disadvantage. This hurdle ...