Skip to main content
Search

# Maximum Mean Discrepancy

3 related articles

Subscribe to updates
  1. arXiv

    A quantum generative model's Fisher trace is pinned to its parameter count, so certification shifts to a two-metric escalation rule

    For quantum circuits that prepare real-amplitude states from half-angle rotations, the classical Fisher information matrix equals the quantum one off a measure-zero set, so its trace is fixed at the parameter count 2n−1 and says nothing about fit quality. The analysis turns to the Fisher eigenvalue rank and to two distributional metrics, Wasserstein-1 and maximum mean discrepancy: the rank only counts the support of the converged distribution and falls to chance on a full-support benchmark. For two analytically solvable bitstring defect families, the tolerance-mismatch window in which the two metrics disagree opens only at certification-grade tolerances and closes as the qubit count grows.

  2. arXiv

    A deep learning surrogate optimizes polycrystalline copper microstructures under competing strength requirements, raising spall strength by up to about 29% with prediction errors below 3%

    The work builds a deep learning-driven framework for optimizing polycrystalline microstructures: a 3D U-Net surrogate maps polycrystalline copper microstructures directly to full-field velocity histories from plate-impact simulations, and is coupled with stochastic microstructure generation and derivative-free optimization to maximize spall strength under a yield strength constraint, achieving up to about a 29% increase in spall strength with CPFE-validated errors below 3% and roughly four orders of magnitude lower cumulative evaluation time than direct CPFE-based optimization.

  3. arXiv

    Post-training diffusion language models with MMD in frozen feature space: MDLM-MMD and ELF-MMD lower generative perplexity at matched entropy, and 16B DMax raises decoding parallelism on math and code benchmarks

    The work introduces a post-training method for diffusion language models that minimizes Maximum Mean Discrepancy (MMD) between generated and reference distributions in the contextual token-feature space of a frozen pretrained DLM, optimizing discrete models with REINFORCE plus a leave-one-out baseline and continuous models by differentiating through generated latents, so that no full sampling trajectories or jointly trained auxiliary models are needed; it reports lower generative perplexity at comparable entropy on OpenWebText, better accuracy-computation trade-offs on GSM8K, and increased decoding parallelism at similar or higher accuracy on 16B DMax-LLaDA2.0 hybrid masked-uniform diffusion models.

You’re all caught up