sidereal

Optimal margin classifier has constraint: $y^{(i)}(w^Tx^{(i)}+b) >= 1$L1 regularization form has constraint: $y^{(i)}(w^Tx^{(i)}+b) >= 1- \zeta_i$These constraints are the condition of not causing training error.It means that when these constraints are satiesfied, there wouldn't be an erro Read More

feature mapping $\phi : R^n \rightarrow R^t$replace $x^{(i)}$ with $\phi(x^{(i)})$since $X=[x^{(i)T}] \epsilon R^{m*1}$, $\phi = [\phi(x^{(i)T})] \epsilon R^{m*1}$$K_{ij}=\phi(x^{(i)})^T\phi(x^{(j)}) \epsilon R^{m*m} = \phi \phi^T \epsilon R^{m*m}$ Read More