Incorporating Reinforced Adversarial Learning in Autoregressive Image Generation

doi:doi:10.57702/7vvjcf89

Incorporating Reinforced Adversarial Learning in Autoregressive Image Generation

Autoregressive models recently achieved comparable results versus state-of-the-art Generative Adversarial Networks (GANs) with the help of Vector Quantized Variational AutoEncoders (VQ-VAE). However, autoregressive models have several limitations such as exposure bias and their training objective does not guarantee visual fidelity. To address these limitations, we propose to use Reinforced Adversarial Learning (RAL) based on policy gradient optimization for autoregressive models.

BibTex:

@dataset{Kenan_E_Ak1_and_Ning_Xu2_and_Zhe_Lin2_and_Yilin_Wang2_2024,
    abstract = {Autoregressive models recently achieved comparable results versus state-of-the-art Generative Adversarial Networks (GANs) with the help of Vector Quantized Variational AutoEncoders (VQ-VAE). However, autoregressive models have several limitations such as exposure bias and their training objective does not guarantee visual fidelity. To address these limitations, we propose to use Reinforced Adversarial Learning (RAL) based on policy gradient optimization for autoregressive models.},
    author = {Kenan E. Ak1 and Ning Xu2 and Zhe Lin2 and Yilin Wang2},
    doi = {10.57702/7vvjcf89},
    institution = {No Organization},
    keyword = {'Autoregressive Models', 'Generative Adversarial Networks', 'Reinforced Adversarial Learning', 'Vector Quantized Variational AutoEncoders'},
    month = {dec},
    publisher = {TIB},
    title = {Incorporating Reinforced Adversarial Learning in Autoregressive Image Generation},
    url = {https://service.tib.eu/ldmservice/dataset/incorporating-reinforced-adversarial-learning-in-autoregressive-image-generation},
    year = {2024}
}