← Back to glossary
Browse letter O hub
Overfitting
A machine learning problem where a model learns the training data too closely and performs poorly on new or unseen inputs.
Definition
A machine learning problem where a model learns the training data too closely and performs poorly on new or unseen inputs.
How It Works
Overfitting helps teams build predictable AI and translation workflows by setting clear expectations for quality, consistency, and decision-making.
In production environments, this concept is applied with process controls such as human review, terminology alignment, and repeatable quality checks across multilingual content.
Key Concepts
- core principle of overfitting
- workflow-level implementation
- terminology and quality consistency
- human validation before publication
Where It Is Used
- localisation workflows
- AI translation pipelines
- multilingual content production
- cross-referencing related concepts such as OCR (Optical Character Recognition)