N-Grams of 5 million books


I watched a cool and short video about n-grams created from 5 million books by using Google’s computational power.Here is a little background info about N-gram: http://en.wikipedia.org/wiki/N-gramN-grams are used in natural language processing (NLP) a lot. We can basically say that they are mostly used to guess the next word according to the previous word sequence.[youtube]http://www.youtube.com/watch?v=5l4cA8zSreQ[/youtube]Here is the talk pageGoogle’s N-gram viewer: https://books.google.com/ngrams/  

Increase productivity on Mac OS with Alfred


I’m already productive on Mac OS :P If you are big fan of spotlight and ava find, alfred is just for you. I like using keyboard and my shortcuts. Alfred embraces this usability with snappy and sleek UI to increase productivity on Mac OS.It works similar to spotlight but it’s highly customisable. You can change the UI colors, indexing features, results structure etc. In addition to these nice features it also provides quick web search options plus giving you chance to create your custom searches which is the coolest feature in my opinion.Here is the best part, Alfred is a free application for Mac OS, you can get powerpack for 17.To try the free version of Alfred

Ubuntu USB mouse scrolling problem workaround


In Ubuntu 13.10, If you are using desktop based viewport switching which basically makes you change virtual desktops by scrolling to switch to next/previous desktop, you will possibly see this bug. Some of the core ubuntu apps are not getting scroll events if you are using a usb mouse. On the other hand, touchpad works just fine.Here are some of the buggy apps which are affected by ubuntu usb mouse scrolling problem: nautilus gedit synapticName: Regression: Enabling typical bindings in “Desktop-based Viewport Switching” breaks scrollwheel scrolling in some windows with a usb mouse on a laptopHere is the bug: https://bugs.launchpad.net/compiz/+bug/1200829What I understand is that it’s related to gtk and the only way to solve is disabling desktop based viewport switching from compizconfig settings manager. To install it:sudo apt-get install compizconfig-settings-managerLet’s wait for next ubuntu update.PS: I have just updated my ubuntu 13.10, and still no fix of the bug. 

MapReduce explained in one pic


Google’s paper: MapReduce: Simplied Data Processing on Large ClustersGoogle IO 2011, mr on app engine: http://www.youtube.com/watch?v=EIxelKcyCC0Mr with python

How to run Weka on large data sets without getting memory heap errors?


I guess Weka is the official tool of each Msc in CS student.I was trying to work with Weka on a relatively large data, ~50k rows, it gave me this java heap error. The trick is running weka with an additional flag.By default its memory is less than 128mb I guess, I was generous and gave 1024mb, you can tweak it according to your computer’s memory plus your task’s consumptionjava -jar weka.jar -Xmx1024m

My notes - Python tutorial for beginners


I found my notes from the time when I was learning python. They are not quite tidy but might be useful. So I created this small python tutorial for newbies.Python is a hybrid language which is interpreted by interpreter in the shell (or cmd). it has its own shell. You can easily switch to python shell from terminal by just writing “python”.Python is a scripting language just like bash. It is high level and functional. This makes it readable. Nearly every job has a function in python. You can handle so many things in one line.Python scripts can be run from python shell and they can be also run from python modules which are files with "py" extension. Every line in the module is interpreted and run by python shell and if there is an error, the operation stops. If the error is in a condition which is not reachable during that runtime, this error won't be seen by python shell. That's an interesting feature of python. Open a new file. Name it. For ex: hello and Change its extension to "py...

Dataset repositories for academic research


I found a couple of nice and comprehensive ones. If you have any, just share it here. UCI repositories: http://archive.ics.uci.edu/ml/ Infochimps: http://www.infochimps.com New York university: http://pages.stern.nyu.edu/~adamodar/pc/datasets/ The Office of National Statistics in the UK:  http://www.statistics.gov.uk/hub/people-places/ Machine learning dataset repository: http://mldata.org Computer vision/Image processing: http://www.imageprocessingplace.com/root_files_V3/image_databases.htm 

Installing Weka on Mac OS X Lion


I tried to install weka which is a great tool for using basic machine learning algorithm. However, every time I tried to install it gave me an error like this"weka-3-7-9 is damaged and can't be opened." error.I tried with earlier versions, again same error with different version number.At the end after some googling I realized that this is new Mac security stuff. Mac OS is not allowing this app to be installed into computer due to not being from an approved developer.To solve it go to Setting -> Security & Privacy, mark “Allow applications downloaded from” “Anywhere”.Weka dmg can be found here:http://sourceforge.net/projects/weka/ 

Back in the game - Blog tasimaca cilesi


 Upuzun bir aradan sonra blogculuk zanaatina donmus bulunuyorum. Bussuru bussuru yazilcak post listem var. Blog tasima konusunda birkac sacma sapan temel bilgiyi paylasacagim. Bir isi yapan adam sayisi cogaldikca o isle ilgili degerli tamamlayici bilgi miktari da azaliyor. Herkes ayni seyi defalarca yaziyor ve her yazan da insanlarin giris seviyesinde bisileri bilmedigini varsaymiyor.Wordpress guncelleme sikintilariWordpress yapisi geregi php karakteristiklerini cok guzel gostermekte. Eski teknoloji olmasindan mutevellit yeni teknolojilerdeki o stabilite mumla araniyor. Isin ilginc yani ise hem joomla hem de wordpress’te hicbir sey degistirmedigim halde site calismamaya basliyordu. Ve yine isin ilginci birkac gun siteyi kendi haline biraktigimda yeniden calisir duruma geliyordu. Yok yok hosting problemlerinden bahsetmiyorum. Onlar ayri hikaye.Neyse wordpress’e her ne kadar muthis bi sevgi beslesem de bu legacy backendi beni bitiriyor. Onceki blogda her guncellemede sorun cikardi. Ek...

İnsan Değirmeni


Gönül bugday tanesine benziyor,Bizse değirmene,Değirmen nereden bilecek Bu dönüşün hikmeti ne?Değirmen taşına benziyor bedenDüşünce ve kaygı onun suyuSu hep onu dinledi.Taş başından geçeni söyledi.Düşünce ve kaygı suyuyla dönen insan değirmeni... Mevlana


subscribe via RSS