Asked about Memorising instead of learning

What is dropout and why does it work?

An answer that works

Randomly ignoring a fraction of units during training, so the network cannot depend on any single one and must learn redundant representations. It behaves like training an ensemble of sub-networks that share weights.

Reading an answer is not being able to give one. The section that teaches this is Memorising instead of learning, 5 hours of Overfitting & regularisation.

Learn this properly

Also asked about Overfitting & regularisation

How do you know a model is overfitting?Your learning curves have converged and both are poor. What now?Your model scores 0.95 AUC in cross-validation and 0.70 in production. What might have happened?What is the difference between L1 and L2 regularisation?