Noise is any irrelevant, erroneous, or random variation in data that obscures the signal a model is intended to learn. In machine learning, noise can arise from labeling mistakes, measurement errors, missing values, or inconsistent data collection.
Broadly speaking, anything that obscures the signal in a dataset. Noise
can be introduced into data in a variety of ways. For example:
- Human raters make mistakes in labeling.
- Humans and instruments mis-record or omit feature values.