$3.$ You are given a dataset of $10$ x $10$ grayscale images. Your goal is to build a $5-$class classifier. You have to DO the following two options:

$($a$)$ the input is flattened into a $100-$dimensional vector, followed by a fully$-$connected layer with $5$ neurons

$($b$)$ the input is directly given to a convolutional layer with five $10$ x $10$ filters

Can the two approaches achieve similar training results? If so$,$ which one you would choose and why?