Problem #$1$:

$(1$ point$)$ Suppose we are given observations $\{($xi

$,$ yi$)\}$

n

i$=1$ with n $=10,000,$ and we

would like to classify these observations into two classes. However, these observations

are grouped into $100$ sets, each containing $100$ observations. The individual labels yi

for i $=1,...,$ n are unknown. Instead, we are provided with the proportion of positive

labels in each group. Additionally, the feature space is high$-$dimensional, with each

xi in R$500.$ Answer the following questions:

$$ Is it feasible to perform classification given the above training data constraints?

$$ If feasible, design a classifier based on SVM for this purpose.Problem #$1$:

$(1$ point$)$ Suppose we are given observations $\{(x_i,y_i){\}}_i=1^n$ with $n=10,000,$ and we

would like to classify these observations into two classes. However, these observations

are grouped into $100$ sets, each containing $100$ observations. The individual labels $y_i$

for $i=1,$dots,$n$ are unknown. Instead, we are provided with the proportion of positive

labels in each group. Additionally, the feature space is high$-$dimensional, with each

$x_{i}in{R}^{500}.$ Answer the following questions:

Is it feasible to perform classification given the above training data constraints?

If feasible, design a classifier based on SVM for this purpose.