We introduce a novel training principle for probabilistic models that is an alternative to maximum likelihood. The proposed Generative Stochastic Networks (GSN) framework is based on learning the transition operator of a Markov chain whose stationary distribution estimates the data distribution. Because the transition distribution is a conditional distribution generally involving a small move, it has fewer dominant modes, being unimodal in the limit of small moves. Thus, it is easier to learn, more like learning to perform supervised function approximation, with gradients that can be obtained by back-propagation. The theorems provided here generalize recent work on the probabilistic interpretation of denoising auto-encoders and provide an interesting justification for dependency networks and generalized pseudolikelihood (along with defining an appropriate joint distribution and sampling mechanism, even when the conditionals are not consistent). We study how GSNs can be used with missing inputs and can be used to sample subsets of variables given the rest. Successful experiments are conducted, validating these theoretical results, on two image datasets and with a particular architecture that mimics the Deep Boltzmann Machine Gibbs sampler but allows training to proceed with backprop, without the need for layerwise pretraining.
Benchmarking Stochastic Interpolants for Modeling Physical Systems
Department of Mechanical Engineering（机械工程系）；Carnegie Mellon University（卡内基梅隆大学）；Department of Geophysical Sciences Committee on Computational and Applied Mathematics（地球科学系计算与应用数学委员会）；University of Chicago（芝加哥大学）；LMD/IPSL, CNRS, ENS, Université PSL École Polytechnique, Institut Polytechnique de Paris, Sorbonne Université（LMD/IPSL、CNRS、ENS、Université PSL 工程学院、巴黎理工学院、索邦大学）；Réseau de Transport d’Électricité (RTE)（电力传输网络（RTE））
Introduction to Stochastic Differential Equations for Generative Machine Learning: A Variational Perspective
Department of Biology, University of Copenhagen（哥本哈根大学生物系）；Department of Applied Mathematics and Computer Science, Technical University of Denmark（丹麦技术大学应用数学与计算机科学系）；Google DeepMind（谷歌DeepMind）；Technical University of Berlin (TU Berlin)（柏林工业大学）；Technical University of Munich（慕尼黑工业大学）
Generative Modeling via Kernelized Stochastic Interpolants
CCM, Flatiron Institute, New York, USA（CCM，Flatiron研究所，纽约，美国）；Courant Institute of Mathematical Sciences, New York University, New York, USA（数学科学学院，纽约大学，纽约，美国）
SBGD: Improving Graph Diffusion Generative Model via Stochastic Block Diffusion
School of Resources and Environmental Engineering, Hefei University of Technology（资源与环境工程学院，合肥工业大学）；School of Computing and Data Science, University of Hong Kong（计算与数据科学学院，香港大学）
Exploring the Optimal Choice for Generative Processes in Diffusion Models: Ordinary vs Stochastic Differential Equations
Institute of Natural Sciences & School of Mathematical Sciences, Shanghai Jiao Tong University（上海交通大学自然科学研究院与数学科学学院）；University of Science and Technology of China（中国科学技术大学）；Suzhou Institute of Advanced Research, University of Science and Technology of China（中国科学技术大学苏州高等研究院）；City University of Hong Kong（香港城市大学）
ProGen: Revisiting Probabilistic Spatial-Temporal Time Series Forecasting from a Continuous Generative Perspective Using Stochastic Differential Equations
Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））
Stochastic Parameterization of Column Physics using Generative Adversarial Networks
Los Alamos National Laboratory（洛斯阿拉莫斯国家实验室）
我们展示了使用概率机器学习技术来开发大气柱物理的随机参数化。在对NASA的Modern-Era Retrospective analysis for Research and Applications，version 2（MERRA 2）数据进行适当的预处理以最大限度地减少MERRA 2估计垂直速度的高频，高波数分量的影响之后，我们使用生成对抗网络来学习非绝热源垂直剖面的概率分布。这可以被看作是对以前类似但确定性的方法的改进，这些方法试图减轻人类设计的物理参数化的缺点和气候模式中“物理”步骤的计算需求。
Stochastic MeanFlow Policies: One-Step Generative Control with Entropic Mirror Descent
Laboratory for Big Data and Decision（大数据与决策实验室）；National University of Defense Technology（国防科技大学）；Samsung AI Center Cambridge（三星AI研究中心）；Queen Mary University of London（伦敦玛丽女王大学）；Fudan University（复旦大学）；ShanghaiTech University（上海科技大学）；Shanghai Innovation Institute（上海创新研究院）
Xaurora: Generative Weather Forecasting with Denoising Stochastic Interpolants from a Foundation Model Prior
Vrije Universiteit Amsterdam（阿姆斯特丹自由大学）；University of Amsterdam（阿姆斯特丹大学）；Delft University of Technology（代尔夫特理工大学）；Centrum Wiskunde & Informatica（数学与计算机科学中心）
Multimarginal generative modeling with stochastic interpolants
Center for Cosmology and Particle Physics（宇宙学与粒子物理中心）；New York University（纽约大学）；Courant Institute of Mathematical Sciences（库朗数学科学研究所）；University of California, Berkeley（加州大学伯克利分校）
On Conditional Stochastic Interpolation for Generative Nonlinear Sufficient Dimension Reduction
School of Statistics, East China Normal University（东华大学统计学院）；Departments of Data Science and Artificial Intelligence, and Applied Mathematics, The Hong Kong Polytechnic University（香港理工大学数据科学与人工智能系、应用数学系）
Score-Based Generative Modeling through Stochastic Differential Equations
Stanford University（斯坦福大学）；Google Brain（谷歌大脑）
从数据中创建噪声很容易;从噪声中创建数据是生成建模。我们提出了一个随机微分方程（RSDs），通过缓慢注入噪声，将复杂的数据分布平滑地转换为已知的先验分布，以及相应的逆时RSDs，通过缓慢去除噪声，将先验分布转换回数据分布。重要的是，逆时梯度仅取决于扰动数据分布的时间相关梯度场（\aka，分数）。通过利用基于分数的生成建模的进步，我们可以用神经网络准确地估计这些分数，并使用数值解算器来生成样本。我们表明，这个框架封装了以前的方法在基于分数的生成建模和扩散概率建模，允许新的采样程序和新的建模能力。特别是，我们引入了一个预测-校正框架来校正离散逆时模型演化中的错误。我们还推导出一个等价的神经ODE，它从与神经元相同的分布中采样，但另外还可以进行精确的似然计算，并提高采样效率。此外，我们提供了一种新的方法来解决逆问题的分数为基础的模型，类条件生成，图像修补和彩色化的实验证明。结合多个架构改进，我们在CIFAR-10上实现了破纪录的无条件图像生成性能，Inception评分为9.89，FID为2.20，竞争可能性为2.99 bits/dim，并首次从基于得分的生成模型中展示了1024 x 1024图像的高保真生成。
Generative Modeling of Neural Dynamics via Latent Stochastic Differential Equations
Department of Machine Learning and Neural Computing, Donders Institute for Brain, Cognition and Behaviour, Radboud University（机器学习与神经计算系，脑认知与行为研究所，拉德堡德大学）
