{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/activation-maximization-generative","title":"Activation Maximization Generative Adversarial Nets","arxiv_id":"1703.02000","date":"2017-03-06","proceeding":"ICLR 2018 1","authors":["Zhiming Zhou","Han Cai","Shu Rong","Yuxuan Song","Kan Ren","Wei-Nan Zhang","Yong Yu","Jun Wang"],"abstract":"Class labels have been empirically shown useful in improving the sample\nquality of generative adversarial nets (GANs). In this paper, we mathematically\nstudy the properties of the current variants of GANs that make use of class\nlabel information. With class aware gradient and cross-entropy decomposition,\nwe reveal how class labels and associated losses influence GAN's training.\nBased on that, we propose Activation Maximization Generative Adversarial\nNetworks (AM-GAN) as an advanced solution. Comprehensive experiments have been\nconducted to validate our analysis and evaluate the effectiveness of our\nsolution, where AM-GAN outperforms other strong baselines and achieves\nstate-of-the-art Inception Score (8.91) on CIFAR-10. In addition, we\ndemonstrate that, with the Inception ImageNet classifier, Inception Score\nmainly tracks the diversity of the generator, and there is, however, no\nreliable evidence that it can reflect the true sample quality. We thus propose\na new metric, called AM Score, to provide a more accurate estimation of the\nsample quality. Our proposed model also outperforms the baseline methods in the\nnew metric.","url_abs":"http://arxiv.org/abs/1703.02000v9","url_pdf":"http://arxiv.org/pdf/1703.02000v9.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"activation-maximization-generative","repo_url":"https://github.com/ZhimingZhou/AM-GAN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"activation-maximization-generative","repo_url":"https://github.com/ZhimingZhou/AM-GAN2","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"diversity","task_name":"Diversity"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.02000","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}