Sunday, November 14, 2010

Bayes' theorem : A Love Story

A few days ago, I saw the video of Hilary Mason presenting the history of machine learning. As far as I know, she has covered the significant developments in the field in the most simplistic, understandable and casual way possible. She shed some light on the fundamental math and algorithmic tools employed in the field.

Hilary suggested, that after the AI winter storm, the best thing that happened in computer science was when computer scientists started invading the field of statistics. That is they starting using probability theories to built application and solve problems. She went on to give an example of how Bayes Theorem was used (and can be used) to classify text, detect spam, etc. I can very well identify with her point.

When I started my MS in computer science, I was supposed to take courses like Software Design, Advance DB, Advance Operating in my first semester. I couldn't get in the Advance Operating Sys class and so I enrolled in Data Mining class. The data mining techniques used, the algorithms that solve complex problems and simply finding interesting patterns in raw data just blew me away. Needless to say, taking DM was probably the best decision of my academic career. I changed my concentration from "Software Engineering" to "Machine Intelligence", worked on making application become smarter (rather than just building a smart application to pass a course) and read more about interesting concepts than that was covered in the class.

My first DM project was building a Spam Filter using Bayes' Theorem. Even though I built much complex DM applications after that, it remains one of my favorite things that I have done. I remember that project presentation was an option to get extra credit (project executable and report however accounted for 40% of the grade). I was doing pretty well in that class but I really wanted to present my Spam Filter. I have never felt so excited about implementing and presenting a project, so I knew that this is something special. And hence, even today when I take on projects that require use of more sophisticated algorithms like SVM, etc. I still perform a quick test using Bayes' Theorem just to get some insight. Maybe it is not a correct approach but I have felt that it has helped me fine tune the algorithms I have to use.

Being a big fan of all things Bayes, I wanted the undergrad and grad students in my college to be excited about using statistics in their projects as well. And hence I gave short tutorial about creating a spam filter using naive bayes theorem during one of the ACM open house meetings. I did that because, I realized that not every CS student enrolled in data mining. The meeting was pretty successful and I tend to believe that I at least made a handful of students excited about using Bayes' theorem.

So if you are an undergard (or even grad) and you are thinking about what project to work on, I'll suggest that you look up some machine learning techniques and see how you can use them to make applications related to your course. There is no better feeling than seeing your application run through mountains of raw(unstructured) data and then providing you with amazing insights and solution.

13 comments:

Anonymous said...

Thanks for this worthy article, it's absolutely acclaimed blogs

Anonymous said...

Hi, і feel that i nоticed you visited
my wеb site sο i came to gо back the favor?
.I'm trying to find issues to improve my web site!I guess its adequate to make use of some of your concepts!!

Here is my web blog :: http://www.allplan-Usa.com
Also visit my blog post home laser hair

Anonymous said...

Wοw that was unuѕual. I just wrotе an extremelу long commеnt but after I clіcked submit my comment didn't show up. Grrrr... well I'm not writing аll that ovеr again.

Аnywayѕ, just wantеd to say ωonderful
blog!
Take a look at my page - V2 Cigs Review

Anonymous said...

Grеat article! Thаt іs thе κind of infoгmation thаt
aгe meant to be sharеd аcross thе wеb.
Disgrаce on the searсh engіnes for no lοngeг
poѕitioning thiѕ submit upper! Ϲome оn oѵer and discuѕѕ with my sіte .

Thаnkѕ =)

mу page www.Prweb.com

Anonymous said...

What's Taking place i am new to this, I stumbled upon this I have found It absolutely useful and it has helped me out loads. I hope to contribute & help other customers like its helped me. Good job.

Also visit my blog - http://www.prweb.com/

Anonymous said...

Normally Ι ԁo nοt learn artiсle on blogs, howevеr
I wiѕh to ѕay that thiѕ wгitе-up very forced mе to сheck оut аnd ԁo it!
Your writing taѕte haѕ been suгрrised me.
Thаnks, quіte niсе poѕt.



Feеl free to surf to my homepаge .
.. http://www.prweb.com/releases/silkn/sensepilreview/prweb10193901.htm

Anonymous said...

Hi! ӏ just wanted to aѕk if you ever have anу
trouble ωith hackers? Mу last blog (ωоrdpгeѕs) ωas hacκed and I ended uρ lοsing
manу mοnthѕ of hard woгk ԁue to no datа
baсkup. Do you have anу methods tο protect agаіnѕt hackers?


Feеl free to surf to mу site - Buy Silk N Sensepil

Anonymous said...

Prеtty! Тhіs hаѕ beеn a reallу wondeгful article.
Mаny thanκs fοr ргοviԁіng these detaіls.



my webpage - please click the following internet page

Anonymous said...

Do you have a ѕpam problem on this blog; Ι also am
a blogger, anԁ I ωаs wonԁering your situation;
mаny of us hаѵe deνеloped some nіce methods anԁ wе are loоking
to ѕwap tесhniques ωith
others, why not shoot me аn e-mail if іntereѕted.


Here is mу blog ... V2 Cigs review

Anonymous said...

I love that this game has a variety of players; if you go to the lower coins tournaments, you could encounter beginner players.
The 8 ball pool hack 2014 failed to seem to bring about waste materials and also
could not possibly be obtained in forested acres, jungles, streams,
lakes, seas, and so forth. Online games were invented few years ago
and soon it gained huge popularity all around the world.


Here is my weblog - tumblr.com

Anonymous said...

Useful info. Fortunate me I discovered your site by chance, and I am surprised why
this coincidence did not came about earlier!

I bookmarked it.

My homepage :: drain snake Goodyear

oakleyses said...

true religion jeans, ray ban pas cher, michael kors, oakley pas cher, polo lacoste, michael kors outlet online, ray ban uk, mulberry uk, michael kors outlet online, coach outlet, burberry handbags, nike air force, vans pas cher, true religion outlet, michael kors outlet, abercrombie and fitch uk, nike roshe run uk, replica handbags, burberry outlet, michael kors outlet online, kate spade, hollister pas cher, michael kors outlet online, sac hermes, nike air max uk, true religion outlet, nike air max uk, michael kors outlet, guess pas cher, north face, hollister uk, new balance, nike tn, uggs outlet, uggs outlet, ralph lauren uk, michael kors, true religion outlet, nike air max, timberland pas cher, nike blazer pas cher, sac vanessa bruno, converse pas cher, coach purses, coach outlet store online, north face uk, nike free uk, hogan outlet, michael kors outlet

Unknown said...




I have to thank you for the efforts you've put in writing this blog
https://3kary.com/property-status/villas-for-rent/
https://3kary.com/property-status/villas-for-rent/
https://3kary.com/property-status/villas-for-sale/
https://3kary.com/property-type/villas/