IPA IP 2022 Question 24
A question IPA Japan published in 令和4年度 ITパスポート試験 公開問題 問24, translated into English. ITPEC is based on the same exams.
Which of the following is the most appropriate description of an example of supervised learning?
Explanation
Inferring why an infant cries from data of recorded cries paired with their causes — supervised learning learns from examples that come with the correct answer.
In supervised learning every training example carries a label, the answer the model should give. Here each recorded cry (the input) is paired with its known cause (the label, such as hunger or a wet diaper). The model learns the relationship between the sound and the cause, and can then infer the cause for a new cry it has never heard: a classification task.
Answer (d)
Why not others:
- •(a) — the site finds the user's taste from the features of images with no correct answer given; grouping by similar features is the work of unsupervised learning
- •(b) — a rule-based program follows rules people wrote in advance; it learns nothing from data, so it is not machine learning at all
- •(c) — a robot that improves a motion through repeated trials, rewarded when the serving goes well, is reinforcement learning
Key rule: Labelled examples (input plus correct answer) = supervised; no labels, find structure = unsupervised; trial and error with rewards = reinforcement learning.
Basic Theory · AI-written, may contain errors
This question comes from the past questions IPA Japan publishes, translated into English by ITPEC Practice. ITPEC Practice is an independent study tool and is not affiliated with IPA or ITPEC. See the original paper (PDF) or Report an issue.