[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Improved Cube Handling in Races

Posted By: sita
Date: Wednesday, 18 June 2014, at 6:53 p.m.

In Response To: Improved Cube Handling in Races (Axel Reichert)

yes, I used the same set of positions. "Of course I did", I am tempted to write. Would you trust a car manufacturer that claims superior crash test results, but gives you the result of his front crash and compares with side crash results of the competition? Hopefully not.

I'm not sure, whether I get your analogy. Of course I meant to use these new positions to to calculate a new total error for Isight as well as its competitors. And yes, that way I would trust the result more.

I agree that 50,000 is a lot, so over-fitting might be unlikely. But these positions are all the race positions from a set of matches, right? So most of them are not close decisions. Which means that only a fraction of these positions really contributed to the training data for the model. The positions which are not close won't help a lot. So, I won't rule out the existence of over-fitting.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.