香港新浪網 MySinaBlog
« 上一篇 | 下一篇 »
心齋 | 13th Sep 2009, 23:40 PM | 創作是一種感受
That's one hell of an equation
With some silly integration
If I were to look at the question
I would say it has a complex solution

But I have to classify the data
Into their appriopriate class
It is really not about the accuracy
But how long it will last

What? I need to find the eigenvalue?
And search the vector space?
I hope I can find the inverse
Because then I will call it a day

They say you have to find a hyperplane
Where the data separate so well
Or you can use some neual net
But how long it will take I cannot tell

You see, you have to do this in small step
Like the girl you just met
Really, you shouldn't say she is fat
It will overshoot and never come back

Well, well, how about reducing the variables?
The so-called curse of dimensionality
It really takes some mentality
Unless you don't care to wait till eternity

I would suggest you to use some recursion
I bet the filter one is not the best solution
But what about another paper's suggestion?
Well, that's another question.

Why not use some evolution?
And see if it gives any revelation.
After a thousand generations
You may begin to see the separation

But the most important thing is,
What's the deadline for the application?