One-Hot Encoding Machine Learning
noun phrase
Definition: A categorical encoding scheme in which each category is represented by a binary vector, or equivalently by a set of binary columns, with exactly one position set to 1 and all others set to 0; scikit-learn explicitly describes it as a one-hot, “one-of-K,” or “dummy” encoding scheme [scikit-learn OneHotEncoder Documentation].
Example in context: “We replace the dense trainable embedding matrix with a fixed one-hot encoding of the vocabulary as the first and last layers of a standard encoder-decoder model.” [Shaham & Levy 2021]
Synonyms: one-of-K encoding; dummy encoding
Related terms: categorical encoding, OrdinalEncoder