fix: last fixes

report/chapters/7_L09_Support_Vector_Machines.typ

@@ -9,7 +9,7 @@
 #eqcolumns(2)[
   == Linear SVM
 
-  We start by analyzing the performance of the `Linear SVM` model. we notice that the `minimum DCF` if we exclude the case with the regularization term $C = 10^(-5)$, is consistent across our range of $C$ values. The `actual DCF`, on the other hand, is more affected by the regularization term $C$ and is inversely proportional to it.
+  We start by analyzing the performance of the `Linear SVM` model. we notice that the `minimum DCF` if we exclude the case with the regularization term $C = 10^(-5)$, #footnote[No support vector has been found, i.e. $bold(alpha)^*$ = [0, ..., 0]] is consistent across our range of $C$ values. The `actual DCF`, on the other hand, is more affected by the regularization term $C$ and is inversely proportional to it.
 
   The model performs poorly compared to the ones we have analyzed before and remains poorly calibrated even with high values of $C$, the `actual DCF` seems to tend to a value of `0.5` for $C -> infinity$
 

report/chapters/8_L10_Gaussian_Mixture_Models.typ

@@ -53,7 +53,7 @@ Perhaps surprisingly, increasing the number of components for the `True` class i
 
 The diagonal covariance model performs better than the full one and I believe that to be thanks to the fact that we are using multiple Gaussian distributions to model the data. We have seen in @filter-last-two that the `Naïve Bayes` model already performs better than the `Multivariate Gaussian` model without the last two features and these last two features are the ones that most benefit from the `GMM` model.
 
-== Best Models for Compared for Different Applications
+== Best Models Compared for Different Applications
 
 #[
   #set par(justify: false)