arXiv research

Paper proposes a new generative model for discrete distributions using flows on submanifolds.

This work analyzes discrete diffusion models using stochastic integrals, providing error bounds and insights.

Faster sampling in discrete diffusion models with predetermined transition time.

Masking diffusion outperforms other discrete diffusion models by incorporating jump times into the model.

Study proposes Local Linear Encoding for better feature discretization.

Discrete geometry model approximates Willmore energy.

While normalizing flows have led to significant advances in modeling high-dimensional continuous distributions, their applicability to discrete distributions remains unknown. In this paper, we show that flows can in fact be extended to discrete events---and under a simple change-of-variables formula not requiring log-d…

New method distills discrete diffusion models, maintaining quality and diversity.

New discrete models for constant mean curvature surfaces and tori.

Paper introduces DMPMs for efficient discrete data generation with sharp convergence bounds.

New method fine-tunes discrete diffusion models for RLHF tasks.

Corrected samplers reduce discretization error in discrete flow models without additional computational cost.

DFMs enable flow-based models for multimodal discrete and continuous data.

Proposes DAM for optimizing discrete generative models.

Probabilistic models with discrete latent variables naturally capture datasets composed of discrete classes. However, they are difficult to train efficiently, since backpropagation through discrete variables is generally not possible. We present a novel method to train a class of probabilistic models with discrete late…

Discrete diffusion models improve data generation for discrete data like language and graphs.

Estimating causal models from observational data is a crucial task in data analysis. For continuous-valued data, Shimizu et al. have proposed a linear acyclic non-Gaussian model to understand the data generating process, and have shown that their model is identifiable when the number of data is sufficiently large. Howe…

Unified framework for continuous-state discrete flow matching models.

New framework for discrete-state diffusion models reduces sample complexity.

Study approximates financial market with discrete-time models.

Study of discrete-time mean-variance model using reinforcement learning.

Consider power utility maximization of terminal wealth in a 1-dimensional continuous-time exponential Levy model with finite time horizon. We discretize the model by restricting portfolio adjustments to an equidistant discrete time grid. Under minimal assumptions we prove convergence of the optimal discrete-time strate…

A new model for time series using discrete latent states.

This paper analyzes discrete diffusion models, deriving convergence bounds for their generated samples.

Unified framework for discrete diffusion modeling with flexible noising processes.

Unified error analysis for discrete flow models.

A new graphical model for discrete data without parametric restrictions.

Paper analyzes symbolic-dynamics inspired Markov modeling for time-series data.

Discrete Flow Maps bypass sequential prediction limits for parallel text generation.

Framework for training-free guidance in discrete diffusion models for molecular generation.

Exact guidance for discrete data improves posterior sampling efficiency.

Discrete noise improves graph generation quality and speed.

IDF++ improves integer discrete flows for lossless compression.

New method boosts performance of diffusion models on discrete data like natural language.

Time series data that are not measured at regular intervals are commonly discretized as a preprocessing step. For example, data about customer arrival times might be simplified by summing the number of arrivals within hourly intervals, which produces a discrete-time time series that is easier to model. In this abstract…

Graph neural nets improve discrete choice modeling with network effects.

We present a mixed multinomial logit (MNL) model, which leverages the truncated stick-breaking process representation of the Dirichlet process as a flexible nonparametric mixing distribution. The proposed model is a Dirichlet process mixture model and accommodates discrete representations of heterogeneity, like a laten…

The present chapter gives an overview on results for discrete knot energies. These discrete energies are designed to make swift numerical computations and thus open the field to computational methods. Additionally, they provide an independent, geometrically pleasing and consistent discrete model that behaves similarly …

Adaptive discretization improves model-based RL in large spaces.

EDLP samples flat modes in discrete spaces using entropy.

We present a numerical model for the dynamics of thin viscous threads based on a discrete, Lagrangian formulation of the smooth equations. The model makes use of a condensed set of coordinates, called the centerline/spin representation: the kinematical constraints linking the centerline's tangent to the orientation of …

We derive a continuum model for incompatible elasticity as a variational limit of a family of discrete nearest-neighbor elastic models. The discrete models are based on discretizations of a smooth Riemannian manifold $(M,\mathfrak{g})$, endowed with a flat, symmetric connection $\nabla$. The metric $\mathfrak{g}$ deter…

FLDD improves discrete diffusion models by learning a non-Markovian noising process.

Discrete diffusion models improve text and image inference.

Gradient-based MCMC for discrete spaces improves sampling performance.

VADD enhances discrete diffusion models by capturing inter-dimensional correlations, improving sample quality.

The local induction equation, or the binormal flow on space curves is a well-known model of deformation of space curves as it describes the dynamics of vortex filaments, and the complex curvature is governed by the nonlinear Schrödinger equation. In this paper, we present its discrete analogue, namely, a model of defor…

Continuous time framework for discrete data denoising models.