I don't think it makes sense to plot different functions for logistic loss and cross-entropy loss as they are essentially two names for the same thing. Sometimes these names are used to differentiate between an overparametrized (softmax) version vs a non-overparametrized version but that's independent of the loss used. In particular the two formulas that are shown are equivalent (one assumed y is +1, -1 the other assumes y is 0, 1). Showing different graphs for the same formula seems confusing.

The letter S(which is usually stride) hasn't been defined. A description would be useful for many.

I appreciate your work, however there are some major drawbacks. Notation is not clear. You did not explain the mathematical notation ( like %99 of the teachers ) in your cheat sheets, it looks like a textbook therefore they are not useful for the beginners. It is possible that a beginner does not heard anything about transpose of a matrix, gradient, Laplacian, or even what conditional probability is. At least you should note the requirements for the learners.

Thanks

can you add a link to https://github.com/MLEveryday/Machine-Learning-Cheatsheets?
we already finished the first three parts.

In refreshers, "linearly dependence" should be corrected to "linear dependence"

The following:

"Matrix-matrix multiplication – The product of matrices A ∈ Rm×n and B ∈ Rn×p is a
matrix of size Rn×p"

Should say:

"Matrix-matrix multiplication – The product of matrices A ∈ Rm×n and B ∈ Rn×p is a
matrix of size Rm×p"

the first page, prediction value is y, real data should be z

In the definition of Determinant in the VIP Refresher on Linear Algebra and Calculus, it would be much clearer if you add a short expression such as: for a fixed i

i think your cheatsheets is very useful and want translate your cheatsheets to Chinese. is it ok?

4.2 Model selection. Is the "development set" the "validation set"?

First, thanks for these resources!

In VIP Refresher: Linear Algebra and Calculus, it says

Matrix-vector multiplication – The product of matrix A \in R^{m×n} and vector x \in R^{n} is a vector of size R^n, such that:

Shouldn't it be instead:

of size R^m, such that:

The same for Matrix-Matrix multiplication.

Fall 2018 should be translated to "Mùa Thu 2018" not "NGÃ 2018" in cheatsheet-supervised-learning.pdf

Remark: we say that we execute a given policy π if given a state a we take the action a =π(s).

Really cool cheatsheets, I presume you have done using LaTex.

Do you mind sharing LaTex template? It would be great. Thank you

Hello,

I guess the description of the forget gate and the input gate should be swapped in the RNN section.

Hi!

Thanks for putting together this helpful refresher on linear algebra and calculus (link for context)! At the risk of being overly pedantic, I noticed that your definition of the Hessian says:

The hessian of f with respect to x is a n×n symmetric matrix

This is true if f has continuous second partial derivatives, but is not guaranteed in general. (Source)

Can you tell me what tool was used to create the cheatsheets? I really like the style and would like to use the tool for my lectures. Thanks in advance.

I wanted to print the Super VIP Cheatsheet to read and make annotations.
Can you provide the pdf in a more printer-friendly format?
Thanks

there is "Tree-based and ensemble methods" in https://stanford.edu/~shervine/teaching/cs-229/cheatsheet-supervised-learning, but no in pdf

Hi @shervinea , since the Traditional Chinese version of all cheatsheets are finished. How to make the official website shown the Traditional Chinese version？

By the way, the word you can use for "Traditional Chinese" is "繁體中文".

Thanks.

The link in the repo description (the one just under the repo title) is broken. It should point to https://stanford.edu/~shervine/teaching/cs-229/ instead of https://stanford.edu/~shervine/teaching/cs-229.html which 404s at the time of opening this issue. Would save a quick Google search in the future.