The Dirichlet Distribution : Data Science Basics

Рет қаралды 4,926

Күн бұрын

Beta Distribution Video : • The Beta Distribution ...
0:00 Recap of Beta Distribution
2:43 Intro to Dirichlet Distribution
6:01 PDF of Dirichlet Distribution
15:07 Statistics and Convergence

Пікірлер: 14

@Break_down1 3 ай бұрын

Excellent. You are so in touch. Would be interested how you’d present a Dirichlet “process” or Gaussian “process.” And how the latter extends to Gaussian process regression, for example

@Deacc 4 ай бұрын

Golden video 🙌

@ritvikmath 4 ай бұрын

Thanks!

@davidheilbron 5 ай бұрын

Thank you!

@ritvikmath 5 ай бұрын

Welcome!

@Darkev77 5 ай бұрын

My biggest problem with probability distributions is understanding how they came out to be in the first place (derivation from first principles), which is quite hard

@pockysticks5625 4 ай бұрын

Same

@bangtri347 5 ай бұрын

Great video! Btw I recently stumbled on your videos about Reinforcment Learning and I was amazed by the way you explained the concepts. Could you make videos on Q learning, DQN or TDL or other concepts in RL?

@user-oj9iz4vb4q Ай бұрын

The shear quantity of -1's in that formula makes me sad. Especially when the numbers you need to input are all +1 of the quantity you'd expect. Like this could easily be (alpha + k - 1) ! / product(alpha_i!) * p_i ^ alpha_i! where alpha is the total population, alpha_i is each population and k is the number of groups.

@bilalarbouch5849 5 ай бұрын

smote video soon ?

@DarkTobias7 5 ай бұрын

How Dirichlet is useful in unsupervised learning and clustering?

@danielkrasnov3671 5 ай бұрын

A pretty classic use of the Dirichlet in unsupervised learning is a topic model like Latent Dirichlet Allocation. You'll probably need some background in Bayesian statistics but I recommend looking into it if you want to see how this distribution can be used in clustering. Basically since it enforces summation to 1 it's often used when things need membership values for soft clustering.