Amongst the various machine learning algorithms, neural nets have emerged as uniquely powerful and versatile. Their application in an engineering context often differs from traditional usage as there are critical physics constraints that need to be satisfied and well-developed mathematical algorithms we can leverage. Large volumes of data can be used to learn such constraints/procedures, but often such copious data isn’t available in engineering contexts. Even when such data is available, there usually exists important structure in engineering relationships (e.g., differential equations) that we can use to dramatically improve accuracy and/or speed of training. This class is about the emerging field at the intersection of neural nets and engineering.
