# Also in the Article

SVR network
This protocol is extracted from research article:
Real-time liver tracking algorithm based on LSTM and SVR networks for use in surface-guided radiation therapy
Radiat Oncol, Jan 14, 2021;

Procedure

Sets of training and testing samples (Dtrain = {(x1,y1), (x2,y2), …, (xm,ym), yi $∈$ R} and Dtest = {(xm+1,y m+1), (x m+2,y m+2), …, (xm+n,ym+n), yi $∈$ R}, respectively) were obtained for the SVR network, which attempts to find a model f(x) in which y* = f(xi) and yi are as close as possible [54, 61]. For a maximum tolerable deviation between y* and yi of ϵ, the SVR problem can be formalized as

where C is the regularization constant and $lϵ$ is the ϵ-insensitive loss function.

By introducing the slack variables $ξi$ and $ξ^i$, Eq. (10) can be expressed as

The Lagrange multiplier $ui$ can then be introduced to obtain the SVR solution as

where b is the model parameter to be determined and f(x) is the final model found by the SVR method.

Note: The content above has been extracted from a research article, so it may not display correctly.

Q&A