Naive Bayes Classifier

Why It Was Important

Naive Bayes applies Bayes' theorem while pretending each feature contributes independently of the others. That assumption is almost never true, yet the classifier is fast, needs little training data, and is hard to beat on text. It became the standard spam filter of the early web and remains the baseline any new classifier is expected to beat before anyone takes it seriously — the clearest demonstration in machine learning that a wrong model can still be a useful one.

Who Invented It

Melvin Maron, Marvin Minsky

Formalised for document classification in the early 1960s; the underlying rule is Thomas Bayes' 18th-century theorem.

Applications

• Spam Filtering
• Document Classification
• Sentiment Analysis
• Medical Screening