Machine Learning

• Regression with 2-means clustering Annoy (non-scientific)

  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…)

• Regression with Annoy (cont’d)

  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…)

• Quick experiment, boosting vs annoy

  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…)

• Further Hack on the Multi-Dimensional Approximate Neighbour Search

  Traversing a tree structure often involves writing a recursive function. However, Python isn’t the best language for this purpose. Therefore I started flattening the tree into a key-value dictonary structure. Logically it is still a tree, but it is physically stored as a dictionary. Therefore it is now easier to write a simple loop to traverse it.

  (more…)

• Approximate Neighbour Search in Multiple Dimensions

  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.

  (more…)

• Re-implementing Approximate Nearest Neighbour Search

  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.

  (more…)

• Random notes on Pandas and Scikit-learn

  So I first heard about Panda probably a year ago when I was in my previous job. It looked nice, but I didn’t really get the chance to use it. So practically it is a library that makes data looks like a mix of relational database table and excel sheet. It is easy to do query with it, and provides a way to process it fast if you know how to do it properly (no, I don’t, so I cheated).

  (more…)

Random Posts

• Quick Hack: Terminal Appearing from Right

  It is useful to have the terminal around whenever I code. However, while real screen estate is finite, having a terminal further limiting the amount of information that can be displayed at the same time. The problem with the terminal is that I don’t really need it all the time, so I usually find it buried under a group of windows.

  (more…)

• Quick and Dirty way to manage my VIM config

  Not sure about the others, but the obsession to my coding tools is probably more than I would admit. I have just managed to do a dirty quick hack to manage my VIM configuration settings. While I am sure there are other people doing this, I would like to show my reinvented wheels.

  (more…)

• Estimating the value of pi with a line of Python

  I came across a video on Youtube on Pi day. Coincidently it was about estimating the value of Pi produced by Matt Parker aka standupmaths. While I am not quite interested in knowing the best way to estimate Pi, I am quite interested in the algorithm he showed in the video however. Specifically, I am interested to find out how easy it is to implement in Python.

  (more…)

• Understanding potree hierarchy file

  One of my recent tasks involving crawling a lot of geo-tagged data from a given service. The most recent one is crawling files containing a point cloud for a given location. So I began by observing the behavior in the browser. After exporting the list of HTTP requests involved in loading the application, I noticed there are a lot of requests fetching resources with a common rXXX pattern.

  (more…)

• Re-implementing Approximate Nearest Neighbour Search

  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.

  (more…)