paper-with-me

Papers

FFJORD: Free-form Continuous Dynamics for Scalable Reversible Generative Models

2018-10-02 · ICLR 2019 5 · Will Grathwohl, Ricky T. Q. Chen, Jesse Bettencourt, Ilya Sutskever, David Duvenaud

A promising class of generative models maps points from a simple distribution to a complex distribution through an invertible neural network. Likelihood-based training of these models requires restricting their architectures to allow cheap computation of Jacobian determinants. Alternatively, the Jacobian trace can be used if the transformation is specified by an ordinary differential equation. In this paper, we use Hutchinson's trace estimator to give a scalable unbiased estimate of the log-density. The result is a continuous-time invertible generative model with unbiased density estimation and one-pass sampling, while allowing unrestricted neural network architectures. We demonstrate our approach on high-dimensional density estimation, image generation, and variational inference, achieving the state-of-the-art among exact likelihood methods with efficient sampling.

📄 PDF Abstract BibTeX arXiv:1810.01367

Code (7)

BorealisAI/continuous-time-flow-process pytorch
francois-rozet/zuko pytorch
jackgoffinet/ffjord-lite pytorch
jacobjinkelly/easy-neural-ode jax
lenz3000/ffjord-path pytorch
mandubian/pytorch-neural-ode pytorch
rtqichen/ffjord pytorch

Tasks

Density EstimationFormImage GenerationVariational Inference

Similar Papers 제목 키워드 기반

Enhanced Distribution Modelling via Augmented Architectures For Neural ODE Flows

2023-06-05 · Etrit Haxholli, Marco Lorenzi

While the neural ODE formulation of normalizing flows such as in FFJORD enables us to calculate the determinants of free form Jacobians in O(D) time, the flexibility of the transformation underlying neural ODEs has been …

Density Estimation

Gaussianization Flows

2020-03-04 · Chenlin Meng, Yang song, Jiaming Song, Stefano Ermon

Iterative Gaussianization is a fixed-point iteration procedure that can transform any continuous random vector into a Gaussian one. Based on iterative Gaussianization, we propose a new type of normalizing flow model that…

Training-Free Continuous Bitrate Control for Scalable Image Coding for Humans and Machines

2026-05-29 · Yui Tatsumi, Hiroshi Watanabe arxiv

Continuous variable-rate compression is highly demanded in real-world applications, but remains underexplored in scalable image coding for humans and machines. In this paper, we propose a training-free variable-rate scal…

Continuous quantification of viral plaque dynamics using ultra-large-area label-free imaging enables rapid antiviral susceptibility testing

2026-05-03 · Merve Eryilmaz, Yuzhu Li, Xiao Wang, Max Zhang 외 arxiv

The plaque reduction assay (PRA) remains the gold standard for antiviral susceptibility testing, evaluating drug potency by measuring reductions in plaque-forming units (PFUs). However, the traditional PRA is time-consum…

Spiking Neural Networks for Continuous Control via End-to-End Model-Based Learning

2025-09-03 · Justus Huebotter, Pablo Lanillos, Marcel van Gerven, Serge Thill arxiv

Despite recent progress in training spiking neural networks (SNNs) for classification, their application to continuous motor control remains limited. Here, we demonstrate that fully spiking architectures can be trained e…

Continuous Control