Suppose we estimate the regression coefficients in a linear regression model by minimizing begin inline style sum from i equals 1 to n of end style open parentheses y subscript i minus beta subscript 0 minus sum from j equals 1 to p of beta subscript j x subscript i j end subscript close parentheses squared subject to begin inline style sum from j equals 1 to p of end style vertical line beta subscript j vertical line less or equal than s for a particular value of s. As we increase s from 0, the training RSS will .