— A symmetric and finite method for measuring the similarity between two probability distributions.
The Jensen–Shannon divergence (JSD) is a method for quantifying the similarity between two or more probability distributions. Unlike the Kullback–Leibler (KL) divergence, upon which it is based, JSD possesses two crucial properties: it is symmetric, meaning the divergence from A to B is the same as from B to A, and it always yields a finite value. These characteristics make it a more robust and widely applicable measure in various computational fields. It finds extensive use in machine learning for comparing model outputs, in bioinformatics for analyzing genetic sequences, and in natural language processing for text analysis, providing a stable metric for assessing distributional differences.