information would be too small with only star student project one unit overlap, but yet to not have
redundant computation would be too large, with units or over overlap.Therefore, the convolution
kernel of size star student project madurai is chosen in this network. Padding the input making it larger
so that there are student projects in madurai feature units centered on the border did not improve
performance significantly. With no padding, a sub sampling of star M.E student projects and a kernel
size , each convolution layer reduces the feature size from n.