Notes on codes, projects and everything
So apparently Annoy is now splitting points by using the centroids of 2 means clustering. It is claimed that it provides better results for ANN search, however, how does this impact regression? Purely out of curiosity, I plugged a new point splitting function and generated a new set of points.(more…)
After a year and half, a lot of things changed, and annoy also changed the splitting strategy too. However, I always wanted to do a proper follow up to the original post, where I compared boosting to Annoy. I still remember the reason I started that (flawed) experiment was because I found boosting easy.(more…)
While following through the Statistical Learning course, I came across this part on doing regression with boosting. Then reading through the material, and going through it makes me wonder, the same method may be adapted to Erik Bernhardsson‘s annoy algorithm.(more…)
In the previous post, I re-implemented Annoy in 2D with some linear algebra maths. Then I spent some time going through some tutorial on vectors, and expanded the script to handle data in 3D and more. So instead of finding gradient, the perpendicular line in the middle of two points, I construct a plane, and find the distance between it and points to construct the tree.
Recently I switched my search code to Annoy because the input dataset is huge (7.5mil records with 20k dictionary count). It wasn’t without issues though, however I would probably talk about it next time. In order to figure out what each parameters meant, I spent some time watching through the talk given by the author @fulhack.
Implementing a Information Retrieval system is a fun thing to do. However, doing it efficiently is not (at least to me). So my first few attempts didn’t really end well (mostly uses just Go/golang with some bash tricks here and there, with or without a database). Then I jumped back to Python, which I am more familiar with and was very surprised with all the options available. So I started with Pandas and Scikit-learn combo.
This update took me quite a bit more time than I initially expected. Anyway, I have done some refactoring work to the original code, and thought it would be nice to document the changes. Overall, most of the changes involved the refactoring of function names. I am not sure if this would stick, but I am quite satisfied for now.
Back then when I was attending a job interview, I was asked to write a Fizz Buzz program to prove that my coding ability. There was only a pen and a piece of paper, so basically means there’s no way I can refer to the documentation for the API syntax. Fortunately I somehow managed to remember and not screw up.
Should have done this earlier, I was just being lazy to go through all the steps to publish it properly. So here it is, the full source is published to bitbucket. Feel free to fork the project if you are interested. I have not attach a licence to it but it will most probably be BSD licence. I have also uploaded the latest 0.0.2 release to bitbucket and would update the download link posted previously soon.
Recently I volunteered in building a site that reports whether certain websites are blocked locally (please don’t ask why that is happening). As it is a very simple app reporting status I wanted it to be easily scrape-able. One of the decision made was I want it to have things to see on first load, this practically removes the possibility of using react, which is my current favorite.