Flow-based Deep Generative Models
Generative models, such as GANs and VAEs, are used to create new data that resembles existing data. However, these models do not explicitly learn the probability density function of real data, which is a difficult task. A new type of generative model called flow-based deep generative models attempts to address this issue by transforming the data into a more manageable form. This allows for the calculation of the probability density function, making it easier to understand and
Generative models, such as GANs and VAEs, are used to create new data that resembles existing data. However, these models do not explicitly learn the probability density function of real data, which is a difficult task. A new type of generative model called flow-based deep generative models attempts to address this issue by transforming the data into a more manageable form. This allows for the calculation of the probability density function, making it easier to understand and work with.
---
Why it matters: This matters because being able to calculate the probability density function of real data can improve the accuracy and reliability of generative models in applications such as image and audio synthesis.
Source: https://lilianweng.github.io/posts/2018-10-13-flow-models/
This article was originally published at: https://lilianweng.github.io/posts/2018-10-13-flow-models/