Why do smart phones sometimes struggle to recognize new faces?
Deep learning
Why do smart phones sometimes struggle to recognize new faces?
Imagine you're using your phone to unlock it with facial recognition, but it keeps failing even though you've never used it before.
The phone's facial recognition system is trained on many faces, but it forgets how to see new ones. Regularization helps by simplifying the model, so it doesn't become too good at memorizing every face it sees, making it better at recognizing new ones.
Example
If the phone's facial recognition system was like a student who memorized all the answers to a test but couldn't solve new problems, regularization would be like giving the student practice with simpler questions to improve their problem-solving skills.
Remember this
Regularization prevents the facial recognition model from memorizing too much, helping it generalize better to new faces.
Text adapted from Wikipedia, licensed under CC BY-SA 4.0.
Related concepts
Convolutional neural network
Can a neural network learn too well?
dropout works as regularization: it approximates an ensemble of subnetworks
Why does turning off neurons randomly help a brainy computer learn better?
batch size affects generalization: larger batches find sharper minima
Larger batch sizes lead to sharper minima, enhancing generalization by providing more accurate gradient estimates
data augmentation does for generalization: artificially expands training set
How can you teach a computer to see better?
Ridge regression
Why do straight lines sometimes fail to fit our data perfectly?
Unsupervised learning
How can we find hidden patterns in massive, unlabeled data?
Swipe through more Machine Learning concepts
Open Pocket Polymath