The differences in Scikit-Learn, Keras, Pytorch and Tensorflow

The first machine learning library I use is Scikit-Learn, and now, primarily Pytorch. However, I have little impression of using Keras, so I wonder what’s the differences.

Pytorch and Tensorflow

I guess those two libraries are well-known in the the AI industry since deep learning models have exponentially grown in the last decade. These two libraries are both low-level deep learning libraries. I am not sure which one is more popular.

I heard that Pytorch is more popular among researchers, while Tensorflow is more often used in production. I think there are probably two reasons. First, Tensorflow developed by Google is released in 2015, which is two years earlier than Pytorch releasing in 2017. Second, the deployment tools for Tensorflow based models are handy, on the other hand, Pytorch models have to be deployed with more effort on python backend like Django.

In terms of Pytorch, the popularity among research community might due to the simplicity of creating models. I have read some Tensorflow 1.0 codes, it was not easy to read. Tensorflow 2.0 has improved greatly in this perspective. But, anyway, if you are working in research community, go for Pytorch; if you want to land a job in industry, head for Tensorflow.