课程: Introduction to Generative Adversarial Networks (GANs)

Introduction to GANs

课程: Introduction to Generative Adversarial Networks (GANs)

开始一个月试用 购买团队方案

Introduction to GANs

“

- Generative adversarial networks. So there's quite a bit to unpack there. So let's go word by word. So generative. What is generation? Basically means being able to learn from a data set and then create new images from that learned data set. It doesn't mean creating a copy but it means getting inspiration and style from the previous content. Adversarial, that's quite a heavy word. What that actually means is it talks about how the different functions of the system work together and against each other to get the best outcome. And we'll go through that in the architecture section. And networks, that basically is a neural network. So we have multiple two neural networks in this overall system that will work together and against each other to generate this media. Let's have a look in detail. We know it's generated by machines, but what exactly does that mean? What is being generated? How is it being generated? We know it has something to do with machine learning but how does ML work in this context? Usually ML is about learning from data. It's about pattern matching and extrapolation. It's also about summarizing data into an object and using that for inference or predictions. But how does the term prediction actually work with generation? How do all these concepts fit together? So over the next few videos, we'll go through the research, the architecture, the design, and the actual code of how gans work.

内容