Convolutional Coast: CNN Spatial Filtering

Bordering the high-dimensional matrix lies Convolutional Coast. Spatial visual signals are encoded in dense 2D pixel grids.

APEX MISSION OBJECTIVES:

Configure a 3x3 vertical Sobel kernel matrix [[-1,0,1],[-2,0,2],[-1,0,1]].
Slide the kernel window across the 8x8 input grid to compute spatial dot products.
Achieve spatial edge detection sensitivity ≥ 85% to earn Apex Visionary rank.

THEORY NOTE: 2D CONVOLUTION & MAX POOLING

Convolutional layers slide a small weight matrix (Kernel) across spatial inputs to calculate feature maps. Max Pooling downsamples resolution while preserving peak spatial activations.

S(i,j) = (I * K)(i,j) = Σ_m Σ_n I(i+m, j+n) · K(m,n)

CNN 2D SPATIAL KERNEL FILTER & POOLING

Isolate vertical edge features in high-dimensional spatial grids

Presets:

Input Signal (8x8)

220

3x3 Kernel Matrix (K)

Edge Sensitivity:28%

Feature Map (6x6)

220

2x2 Max Pooled (3x3):

220