-
Hi, I would have a couple of more questions on the KernelExplainer:
(1) Are there compelling (analytical) reasons why one should or should not use the full training set as background for the Kernel…
-
Hey there!
I'm working on a project with 9000 samples and 300k features.
These features were used in multi-class classification and regression (multi-output) problems, using a Pytorch neural ne…
-
I am extending all chapters with a section for software implementations and to alternative algorithms (also with software implementation). The software can be any free and open source software: R, …
-
Hi, the kernel explainer is exact to my understanding by summing over all terms in Eq 8 from the paper using Theorem 2. That is, it returns the exact Shapley values, if the background sample is repres…
-
qfield does not assume the relative directory for geopakage.
when charged creates an error and says the layers can not be loaded. I checked the project and it is configured correctly. the problem lie…
1900g updated
5 years ago
-
Hey there! I'm using the gradient explainer to interpret a 8000 sample x 200000 feature array. I split it into training and testing, and subsampled to get 200 train samples and 200 test samples. When …
-
Brilliant work.
_______________________________________________________________________________
Adding LIME advantages and disadvantages.
Explaining what is meant by sparse explanations - Shapley…
-
Shapley values
LocalSurrogate
-
Is it possible to pull files from a sub directory within my GitHub repository.
My stata ado files are organized in the following structure
.../src/a
.../src/e
.../src/_
I would like to gene…
-
Hi @imoscovitz
Right now, I'm training IREP or RIPPER on up to 2 or 3k features.
In the end, the generated rules tend to use only 30 features max.
In my case, doing a such amount of features eng…