Convolution and Feature Maps Playground

Convolutional feature maps are generated through localized matrix dot products traversing a multidimensional input tensor. The operation extracts spatial hierarchies by applying discrete translation-invariant filters. Adjusting stride and padding modifies the resulting tensor dimensions deterministically, following strict algebraic formulas. This mathematically efficient extraction of structural data minimizes parameter space, establishing the computational foundation for sophisticated pattern recognition within deep neural architectures.

A convolution slides a small grid of numbers, the kernel, across the image. At each stop it multiplies the overlapping values and adds them into one output pixel. The result is a feature map that brightens wherever the kernel's pattern appears. Try the edge-detect kernel here: flat regions go dark while boundaries light up, because the kernel responds to change, not to brightness itself.

The operation nearly every CNN calls "convolution" is actually cross-correlation. True mathematical convolution first flips the kernel top-to-bottom and left-to-right before sliding it; deep-learning libraries skip that flip. Since the network learns the weights anyway, a flipped kernel would simply be learned in reverse and give an identical result, which is why the misnomer stuck. With a symmetric kernel like blur, the two are indistinguishable.