Skip to content

Concept Glossary: U

This page lists English glossary entries for this letter. Entry bodies are generated from per-term source files with includes.

uncertainty

  • Meaning: Uncertainty is the condition where the current information does not determine one fixed state or outcome, so multiple possibilities must remain in view.
  • Why it matters: Real problems often require asking how confident a result is, what could still be wrong, and when additional review is needed. Uncertainty connects prediction, probability, risk, human oversight, and automation boundaries.
  • Related concepts: probability, incomplete information, noise
  • Core Section: P1-6.2
  • Appears in: P1-6.1

underfitting

  • Meaning: Underfitting is a state where a model has not learned enough of the relationships in the data, so performance is weak on both training data and new data. The model may be too simple, or training may not have progressed far enough to capture the important pattern.
  • Why it matters: Underfitting helps separate not learned enough from learned the training data too tightly. Without this distinction, a low score may be misread as overfitting even when the model actually needs more capacity, better features, or more effective training. Underfitting is not a desirable form of generality; it is a failure to capture even the basic pattern needed for the task.
  • Related concepts: generalization, overfitting, model complexity
  • Core Section: P1-3.2
  • Appears in: P4-4.1, P4-5.1, P4-5.2

unsupervised learning

  • Meaning: Unsupervised learning is a learning setup that looks for structure, groups, or representations in data without human-provided answer labels. Instead of predicting a predefined target, it first asks how examples resemble or differ from one another.
  • Why it matters: It shows that learning is not only about getting known answers right. Clustering, dimensionality reduction, and representation learning all begin from the question of what structure can be read from the data itself.
  • Related concepts: structure, clustering, dimensionality reduction
  • Core Section: P4-2.2
  • Appears in: P1-8.2, P4-index, P4-2.2, P4-17.1, P4-17.2, P4-18.1, P4-18.2, P4-2.3