Dimensionality Reduction

Category: science

The process of reducing the number of input variables in a dataset to make the model faster and less complex.

Too many inputs can confuse an AI (the "Curse of Dimensionality"). Reduction identifies which of the 100 features actually matter and discards the 80 that don’t. It makes the model faster to train and far more robust to noise.

Common Examples

• We performed dimensionality reduction using PCA to trim our model inputs, cutting training time in half without losing any accuracy.
• Dimensionality reduction prevents models from wasting computational energy on redundant features that add no predictive value.