Case study 3 – using HTDM with prior knowledge of the hyperparameter values
Let’s say, in this case, we are also faced with a similar condition as in the previous case study, but this time, we have prior knowledge of the good hyperparameter values for the given data since one of the data scientists in our team has worked with the same data previously. This means we will only focus on the left branch of the first node in HTDM, as shown here:
Figure 12.10 – Case study 3, have prior knowledge
Based on the given case description, we know that we do not have enough parallel computational resources since we only have a single-core CPU. This means we will only focus on the right branch of the second node, as shown here:
Figure 12.11 – Case study 3, not enough parallel computational resources
We also know that we have a medium-sized hyperparameter space that only consists of numerical types of values. This means...