locked
BPM (Migrated from community.research.microsoft.com) RRS feed

  • Question

  • Ali Yaghoubi posted on 03-31-2011 10:05 AM

    Hi,

    Would you mind telling why in the first BPM example in the tutorial the vector x is augmented by 1?

    Can't we use a 2x2 matrix w instead of a 3x3?

    Thanks

    Friday, June 3, 2011 6:43 PM

Answers

  • John Guiver replied on 04-01-2011 10:04 AM

    This is a constant input whose corresponding weight acts as a bias term to shift decision boundaries away from the origin in input space.

    John

    Friday, June 3, 2011 6:43 PM