After reading the paper and watching the video over and over again, finally I understand how to works.
@user-uv6oh6et4r
6 жыл бұрын
XGBoost - the algorithm that wins every competition.
@geoffreyanderson4719
8 жыл бұрын
I hope Mr. Chen will bring his improvements to more software I use! Mr. Chen did great work on XGBoost, with approximate algorithms, sparsity, out of core, distributed, open, portable, parallel, which are all very good together.
@vinaykrishna1989
5 жыл бұрын
Is it just me who can't follow his words? and caption is doing poor job as well.
@KumarHemjeet
2 жыл бұрын
count me in bro.. :/
@fuhaoda
5 жыл бұрын
In the pruning and regularization slide, there should be 1/2 for the first 3 terms, right?
@rltangyt
7 ай бұрын
You are right. However, there is some flexibility here. If we view gamma as 2 times the original gamma, the formula is still reasonable. Basically, gamma is a tuning parameter so that 2*gamma is still a reasonable tunning parameter.
@maddoo23
2 жыл бұрын
By leaf weight, he means leaf's output, right?
@geoffreyanderson4719
8 жыл бұрын
Can it be possible for GPU accelerator speed, as well as Mr. Chen's other important improvements, also be brought to more learning algorithm implementations, such as XGBoost and others?
Пікірлер: 11