Assume given a dataset $\{($x $($i$),$ y$($i$))$ : i $=1,2,...$N$\}$ where y $($i$)=1$ or $1.$ For these datasets, define the covariance as$,=1$ N $$ X i:y $($i$)=1($x $($i$)0)($x $($i$)0)$ T $+$ X i:y $($i$)=1($x $($i$)1)($x $($i$)1)$ T $$ where $0=$ PN i$=11$y $($i$)=1$x $($i$)$ PN i$=11$y $($i$)=1,1=$ PN i$=11$y $($i$)=1$x $($i$)$ PN i$=11$y $($i$)=1.$ Both LDA and hard$-$margin SVM could generate a linear decision boundary Tx $+$ b$.($a$)(5$ points$)$ Write out and b generated from both two methods, named as LDA$,$ SVM$,$ bLDA, bSVM respectively, besides the parameters we define for you, you can also use the i $,$ which is the Lagrange coefficient of SVM w$.$r$.$t each data $($x $($i$),$ y$($i$)).$ You don$$t need to show your process