Add livebook #262
Add livebook #262
Conversation
|
Looks great! Should we also talk about NNDescent? |
notebooks/nearest_neighbors.livemd
Outdated
| @@ -0,0 +1,255 @@ | |||
| <!-- livebook:{"file_entries":[{"name":"knn_gyms.png","type":"attachment"}]} --> | |||
There was a problem hiding this comment.
Oh, you should also run the notebook with persist outputs enabled, so we have the outputs in the guides. :)
Co-authored-by: José Valim <[email protected]>
Co-authored-by: José Valim <[email protected]>
Co-authored-by: José Valim <[email protected]>
| |  | | ||
| | :------------------------------------------------------------------------------------: | | ||
|
|
||
| Figure 3: A process of creating a tree in Random Projection Forest. Source: http://www.math.umassd.edu/~dyan/rpKNN.png?attredirects=0 |
There was a problem hiding this comment.
Not sure about this figure. The reason is that random projection trees in Scholar always split the data in half. Perhaps the one from https://cseweb.ucsd.edu/~dasgupta/papers/rptree-stoc.pdf or https://cseweb.ucsd.edu/~dasgupta/papers/exactnn-algo.pdf might be a better idea?
There was a problem hiding this comment.
Are there any photos of these on the internet that I could link to?
There was a problem hiding this comment.
I will include the image from the first paper that shows how splitting looks like in the case of kdtree and rpf, but I will leave the original image and mention that actual splitting in Scholar is implemented differently.
There was a problem hiding this comment.
That might be a better idea. :)
There was a problem hiding this comment.
I don't think it's that important, so feel free to leave the existing figure if you want. :)
Co-authored-by: Krsto Proroković <[email protected]>
Closes #237