生成对抗网络GAN的研究进展

Research progress on generative adversarial network

基于零和博弈思想的生成式对抗网络(generative adversarial network,GAN)模型的意义在于可通过无监督学习获得数据的分布,并能生成较逼真的数据。它可以应用在很多领域,尤其是在计算机视觉领域中的图像生成方面取得了很大成果,成为当下的一个研究热点。以GAN模型及其在特定领域的应用结果为研究对象,对GAN的改进和扩展的研究成果进行了广泛的研究,并从图像超分辨率重建、文本合成图片等多个实际应用领域展开讨论,系统地梳理、总结出GAN的优势与不足,同时结合自然语言处理、强化学习对GAN的发展趋势及应用前景进行预测分析。

The significance of the GAN model based on the zero-sum game idea is that the data distribution can be obtained through unsupervised learning,and it can generate more realistic data.It can be applied in many fields,especially the field of computer vision,and achieved great results in image generation,so that it becomes a hot spot in current research.This paper took the GAN model and its application results in specific fields as the research object,conducted extensive research on the improvement and expansion of GAN research results,and discussed from multiple practical application areas such as image super-resolution reconstruction and text synthesis pictures.It systematically sorted out and summarized the advantages and disadvantages of GAN,predicted and analyzed the development trend and application prospect of GAN in combination with natural language processing and reinforcement learning.