Hacker News Reader

I think its because of three reasons:

1. Yandex is annoucing a new ML library and that makes it news because Yandex is well established

2. Gradient boosting is quite effective and popular

3. Not everything has to be about deep learning

Adding to 1.: This came out of MatrixNet research, which was state-of-the-art (and well-guarded) for years.

Deep learning works best on images, video, text, audio and reinforcement learning. Most common applications are outside these domains, based on diverse types of business/scientific data. But, as they said, you can compose CatBoost with TF and Keras neural nets. It connects like a NN module.

Do you have any examples of this or can point me in the right direction of how to implement it? Thanks!

Afaik gradient boosting wins lots of Kaggle competitions when they are on small amounts of data. It is possible that gradient boosting is the king of small data, but I'm not an expert on the subject