A Minimum Variance Path Principle for Accurate and Stable Score-Based Density Ratio Estimation

Researchers propose the Minimum Variance Path (MVP) Principle to improve score-based machine learning methods by addressing the path variance problem that makes theoretically path-independent methods practically path-dependent. The approach uses a closed-form variance expression and Kumaraswamy Mixture Model to learn data-adaptive, low-variance paths, achieving new state-of-the-art results on benchmarks.

• →Score-based methods face a paradox where they are theoretically path-independent but practically path-dependent due to overlooked path variance terms.
• →The MVP Principle provides a closed-form expression for path variance, making optimization tractable for the first time.
• →A flexible Kumaraswamy Mixture Model enables automatic learning of data-adaptive, low-variance paths without manual tuning.
• →The method establishes new state-of-the-art results on challenging benchmarks for density ratio estimation.
• →This framework provides a general approach for optimizing score-based interpolation across machine learning applications.