Weakly Labeled Training Record

A Weakly Labeled Training Record is a training record with a noisy label.

• AKA: Weakly Labeled Training Case.
• Context:
  • It can be a member of a Weakly Labeled Dataset.
  • It can have been labeled by a Labeling Heuristic.
  • …
• Counter-Example(s):
  • a Strongly Labeled Training Record.
• See: Self-Training Algorithm.

---

---

References

2002

• (Bockhorst et al., 2002) ⇒ Joseph Bockhorst, and Mark Craven. (2002). “Exploiting Relations Among Concepts to Acquire Weakly Labeled Training Data.” In: Proceedings of the Nineteenth International Conference on Machine Learning (ICML 2002).

---