Skip to main content Link Menu Expand (external link) Document Search Copy Copied

Weak Supervision

Weak supervision in machine learning refers to the training process where instead of relying solely on fully labeled data, the model is trained using a combination of partially labeled or noisy data, heuristics, or other forms of less precise supervision signals. It allows for training models when obtaining large amounts of accurately labeled data is difficult, expensive, or time-consuming.

Video: