What is the curse of dimensionality and why does Naïve Bayes handle it well?

Naïve Bayes handles high-dimensional data well because it avoids modeling full joint distributions.

In high dimensions, estimating joint probability distributions becomes computationally infeasible due to exponential growth in parameter space. Naïve Bayes avoids this by assuming conditional independence, reducing complexity from O(2^d) to O(d). This makes it particularly effective in text classification, where d can be tens of thousands.