問67
頻出度★★☆
ディープラーニングに関する記述として, 最も適切なものはどれか。
令和4年度 ITパスポート試験 問67
テクノロジ系基礎理論 › 情報に関する理論
This question's explanation and your answers so far.
解説
問 67
Incorrectア
Using teaching materials provided on the Internet to learn according to one's proficiency level, free from constraints of distance and time, describes “e-learning (distance learning)” and has nothing to do with deep learning. Although the names look similar, this is about a mode of learning, not an AI technology.
- イ
正解 イ is correct. Deep learning is a representative method of machine learning: it uses a multi-layered (deeply stacked) neural network to let a computer analyze large amounts of data and, without a human teaching it the features, discover regularities and features in the data on its own, then perform inference and judgment based on the results. The description in choice イ, “analyzes large amounts of data and uses a neural network to discover regularities on its own, performing inference and judgment”, is exactly what this mechanism does, so it is the most appropriate.
Incorrectウ
Providing appropriate answers from systematically classified specialized knowledge in a particular field describes “an expert system”. Because it uses a knowledge base whose rules are defined in advance, its mechanism differs from deep learning, which learns regularities from data on its own.
Incorrectエ
Estimating the difficulty of problems and the knowledge and skills required, according to each individual's proficiency and understanding, describes “adaptive learning”. It is an approach to learning support, not the definition of the deep-learning technology itself.
ここがポイント
- Deep learning is a machine-learning method that uses multi-layered neural networks to automatically learn features and regularities from large amounts of data.
- Distinguishing confusable terms: e-learning (distance learning), expert systems (rule-based use of knowledge), and adaptive learning (individually optimized learning) are all concepts separate from deep learning.
用語