An Interpretable Evaluation of Entropy-based Novelty of Generative Models

Code

• github.com/buyeah1109/KEN
  OfficialIn paperpytorch★ 8

Abstract

The massive developments of generative model frameworks require principled methods for the evaluation of a model's novelty compared to a reference dataset. While the literature has extensively studied the evaluation of the quality, diversity, and generalizability of generative models, the assessment of a model's novelty compared to a reference model has not been adequately explored in the machine learning community. In this work, we focus on the novelty assessment for multi-modal distributions and attempt to address the following differential clustering task: Given samples of a generative model P_G and a reference model P_ref, how can we discover the sample types expressed by P_G more frequently than in P_ref? We introduce a spectral approach to the differential clustering task and propose the Kernel-based Entropic Novelty (KEN) score to quantify the mode-based novelty of P_G with respect to P_ref. We analyze the KEN score for mixture distributions with well-separable components and develop a kernel-based method to compute the KEN score from empirical data. We support the KEN framework by presenting numerical results on synthetic and real image datasets, indicating the framework's effectiveness in detecting novel modes and comparing generative models. The paper's code is available at: www.github.com/buyeah1109/KEN

Tasks

Reproductions