AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Conditional Generation Using Polynomial Expansions

Grigorios Chrysos, Markos Georgopoulos, Yannis Panagakis

NeurIPS 2021 · Image generation › Autoregressive and latent generation · approach: Generative latent space tooling · 2 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace SNGAN class conditioning with CoPE two-variable polynomialSNGAN generator · CIFAR10 class-conditionalFID: 14.70 ˘ 0.97 -> 14.22 ˘ 0.66Improved
replace SNGAN class conditioning with CoPESNGAN generator · CIFAR10 class-conditionalIS: 8.30 ˘ 0.11 -> 8.77 ˘ 0.12Improved
concatenate the class input into SNGAN (SNGAN-CONC)SNGAN generator · CIFAR10 class-conditionalFID: 14.70 ˘ 0.97 -> 30.65 ˘ 3.55Hurt
use SPADE-style conditioning in SNGANSNGAN generator · CIFAR10 class-conditionalFID: 14.70 ˘ 0.97 -> 21.74 ˘ 0.73Hurt
replace Pi-Net with CoPEPi-Net generator · CIFAR10 class-conditionalIS: 7.54 ˘ 0.16 -> 7.87 ˘ 0.21Improved
replace Pi-Net with CoPEPi-Net generator · CIFAR10 class-conditionalFID: 37.26 ˘ 1.86 -> 34.35 ˘ 2.68Improved
replace Pi-Net with CoPEPi-Net generator · Cars196 class-conditionalFID: 120.40 ˘ 28.65 -> 55.48 ˘ 3.16Improved
use products of polynomials (Pi-Net) instead of single-variable concatenation (Pi-Net-SICONC)Pi-Net generator · CIFAR10 class-conditionalFID: 71.81 ˘ 33.00 -> 37.26 ˘ 1.86Improved
replace SPADE with CoPE (adds products of polynomials)Pi-Net generator · Cars196 class-conditionalFID: 168.19 ˘ 39.71 -> 55.48 ˘ 3.16Improved
replace Pi-Net-SICONC with CoPEPi-Net generator · Cars196 super-resolution 8xFID: 186.42 ˘ 40.84 -> 62.76 ˘ 4.37Improved
replace SPADE with SPADE-CoPESPADE · Cars196 super-resolution 16xFID: 111.75 ˘ 13.41 -> 72.63 ˘ 3.18Improved
replace SPADE-CoPE with full two-variable CoPESPADE-CoPE · Cars196 super-resolution 16xFID: 72.63 ˘ 3.18 -> 60.42 ˘ 6.19Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy