REVIEW 5 cited by
Learning Likelihoods with Conditional Normalizing Flows
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Normalizing Flows (NFs) are able to model complicated distributions p(y) with strong inter-dimensional correlations and high multimodality by transforming a simple base density p(z) through an invertible neural network under the change of variables formula. Such behavior is desirable in multivariate structured prediction tasks, where handcrafted per-pixel loss-based methods inadequately capture strong correlations between output dimensions. We present a study of conditional normalizing flows (CNFs), a class of NFs where the base density to output space mapping is conditioned on an input x, to model conditional densities p(y|x). CNFs are efficient in sampling and inference, they can be trained with a likelihood-based objective, and CNFs, being generative flows, do not suffer from mode collapse or training instabilities. We provide an effective method to train continuous CNFs for binary problems and in particular, we apply these CNFs to super-resolution and vessel segmentation tasks demonstrating competitive performance on standard benchmark datasets in terms of likelihood and conventional metrics.
Forward citations
Cited by 5 Pith papers
-
Lyman-$\alpha$ forest holography: 3D predictions from 1D measurements
One-dimensional Lyman-α forest power spectrum measurements, propagated through the ForestFlow emulator, predict three-dimensional clustering that matches DESI BAO and ACCEL-2 simulation results.
-
Flow-Based Surrogates for High-Dimensional Likelihoods in Experimental Neutrino Physics
A hybrid coupling-plus-autoregressive normalizing flow trained on a 110-parameter T2K-like near-detector likelihood reaches 98% relative ESS versus 5% for the post-fit Gaussian and matches MCMC flux predictions.
-
Non-Invasive Reconstruction of Intracranial EEG Across the Deep Temporal Lobe from Scalp EEG based on Conditional Normalizing Flow
A conditional normalizing flow reconstructs band-limited (0.5–50 Hz) intracranial EEG from scalp EEG across multiple medial temporal lobe subregions in three epilepsy patients, with held-out session results comparable...
-
An invertible generative model for forward and inverse problems
A single invertible map constructed from two triangular normalizing flows can conditionally sample both the likelihood and the posterior in Bayesian inverse problems.
-
Generative imaging for radio interferometry with fast uncertainty quantification
RI-GAN couples a regularised conditional GAN with a GU-Net generator to deliver fast radio interferometric image reconstructions and uncertainty maps.
Discussion (0). Sign in to comment.