Research Article
Online Sequential Projection Vector Machine with Adaptive Data Mean Update
| Initial Phase: Given the initial training data , the accumulation ratio . | | (1) Compute the data mean and get ; | | (2) Compute SVD of : ; | | (3) Get the hidden nodes by making ; | | (4) Obtain input weights ; | | (5) Compute the output weights | | Online learning phase: Given the th chunk of data , | | (1) Compute ; | | (2) Compute ; | | (3) Set | | (4) Compute | | (5) Compute ; | | (6) Compute ; | | (7) Compute the QR-decomposition of | | (8) Compute SVD of : | | (9) Get the updated SVD : ; | | (10) Get the new number of hidden nodes by making | | (11) Update input weights ; | | (12) Update the output weights |
|