AI Paper Atlas

Research map / Image generation / GANs

CatGAN: Category-Aware Generative Adversarial Networks with Hierarchical Evolutionary Learning for Category Text Generation

Zhiyue Liu, Jiahai Wang, Zhiwei Liang

AAAI 2020 · Image generation › GANs · approach: GANs for text generation · 79 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

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
Use CatGAN rather than SentiGANcategory text generation · category synthetic, length 20 and 40NLL_oracle: 6.976 -> 6.649 ± 0.097; 6.821 -> 6.498 ± 0.186Improved
Use CatGAN rather than RelGANgeneral text generation · general synthetic, length 20 and 40NLL_oracle: 6.680 -> 6.377 ± 0.116; 6.756 -> 6.235 ± 0.131Improved
Use CatGAN rather than the strongest listed baselinecategory text generation · MRBLEU-4: 0.167 -> 0.194 ± 0.028Improved
Use CatGAN rather than the strongest listed baselinecategory text generation · ARBLEU-4: 0.691 -> 0.867 ± 0.016Improved
Use CatGAN rather than the strongest listed baselinegeneral text generation · ENBLEU-4: 0.501 -> 0.603 ± 0.010Improved
Increase λ from 0 to 1 (ablation)CatGAN · ARNLL_div: 1.470 -> 1.689Improved
Increase λ from 0 to 1 (ablation)CatGAN · ARBLEU-5: 0.757 -> 0.654Hurt
Add hierarchical evolutionary learning (ablation)CatGAN · ARBLEU-5: 0.613 -> 0.751Improved
Add temperature mutation strategy (TMS) (ablation)CatGAN · ARBLEU-5: 0.703 -> 0.751Improved
Add temperature mutation strategy (TMS) (ablation)CatGAN · ARNLL_div: 1.555 -> 1.539No clear effect
Use OMS rather than only the CatRa objective (ablation)CatGAN · ARBLEU-5: 0.638 -> 0.751Improved
Use OMS rather than only the CatRa objective (ablation)CatGAN · ARNLL_div: 1.618 -> 1.539Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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