A Sobel-Gradient MLP Baseline for Handwritten Character Recognition
This work provides a simpler, more interpretable alternative to convolutional neural networks for handwritten character recognition, though it is incremental in scope.
The authors investigated whether first-order edge maps from the Sobel operator could effectively drive a dense multilayer perceptron for handwritten character recognition, achieving 98% accuracy on MNIST digits and 92% on EMNIST letters while offering a smaller memory footprint than CNNs.
We revisit the classical Sobel operator to ask a simple question: Are first-order edge maps sufficient to drive an all-dense multilayer perceptron (MLP) for handwritten character recognition (HCR), as an alternative to convolutional neural networks (CNNs)? Using only horizontal and vertical Sobel derivatives as input, we train an MLP on MNIST and EMNIST Letters. Despite its extreme simplicity, the resulting network reaches 98% accuracy on MNIST digits and 92% on EMNIST letters -- approaching CNNs while offering a smaller memory footprint and transparent features. Our findings highlight that much of the class-discriminative information in handwritten character images is already captured by first-order gradients, making edge-aware MLPs a compelling option for HCR.