论文信息 - DPGAN: PReLU Used in Deep Convolutional Generative Adversarial Networks

DPGAN: PReLU Used in Deep Convolutional Generative Adversarial Networks

This paper is directed against the image super resolution problem that is an extraordinary topic in the field of computer vision. We proposed a new model PReLU Used in Deep Convolutional Generative Adversarial Networks (DPGAN) which has designed a pre-training structure, and the generator and the discriminator are cross-optimized to form a stable network structure. In the proposed model, the activation function in the generator uses the PReLU[1] innovatively. The experimental results demonstrate that the images generated by the proposed model have higher resolution, which is compared with previously studied models.

Xiaofang Liu | Fang Zuo

[1] Yoshua Bengio,et al. Generative Adversarial Nets , 2014, NIPS.

[2] Frank Schorfheide,et al. Loss function‐based evaluation of DSGE models , 2000 .

[3] Tony R. Martinez,et al. The general inefficiency of batch training for gradient descent learning , 2003, Neural Networks.

[4] Simon Osindero,et al. Conditional Generative Adversarial Nets , 2014, ArXiv.

[5] É. Borel. The Theory of Play and Integral Equations with Skew Symmetric Kernels , 1953 .

[6] Soumith Chintala,et al. Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[7] Jian Sun,et al. Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[8] Geoffrey E. Hinton,et al. Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.