Q: 9
Your team is running a simulation-based optimization exercise to increase routing efficiency. Learning
for this exercise is done through "trial and error".
Which type of machine learning approach is being leveraged for this exercise?
Options
Discussion
Not C here. The mention of "trial and error" is a giveaway for B (Reinforcement Learning), since that approach learns by interacting with the environment and getting feedback, not from labeled data sets like Supervised Learning. D is a distractor, because only reinforcement fits that scenario directly. Open to other interpretations if I'm missing something though.
B , trial and error is classic reinforcement learning. C is tempting but that's usually about labeled data so it's a trap option here.
Encountered exactly similar question in my exam, B is what matched there due to the "trial and error" learning pattern.
Is anyone sure A could apply here? "Trial and error" feels pretty specific to Reinforcement Learning, especially since the system learns by interacting with the environment and tweaking for better outcomes. I've seen similar phrasing on other PMI practice sets and it's usually B, but open if there's a good case for A in this routing context.
Hmm, I was thinking A since unsupervised covers scenarios without explicit labels, and simulations sometimes use clustering methods too. B is a strong pick, but A seems plausible if feedback isn't direct. Am I missing something?
C , because if the simulation is using historical routes with known outcomes, that's more like supervised learning. Trial and error could fit B, but if results are labeled up front, C makes sense. I might be off since the wording is vague, anyone think otherwise?
A is wrong, B. Trial and error is basically textbook Reinforcement Learning. You get feedback from the environment after each attempt, which fits exactly with how RL works. Same thing called out in a few practice sets and official guides.
B or maybe A if they didn't say trial and error, but with that phrase it's gotta be B here.
Pretty confident that's B for this. "Trial and error" always screams Reinforcement Learning to me.
B , "trial and error" is classic Reinforcement Learning. The model learns from actions and their rewards in the environment, not from labeled data. Pretty sure that's what the question is pointing at, but open to other views.
Be respectful. No spam.
