aiqbench analysis
This research establishes a controlled baseline by decoupling fixed first-order edge extraction via the Sobel-Feldman operator from learned classification in an MLP, demonstrating that simple gradient representations retain substantial discriminative structure for handwritten character recognition. The study reveals that while this approach achieves near-state-of-the-art accuracy on standard benchmarks, remaining errors are primarily concentrated among geometrically similar classes where edge geometry alone is ambiguous. For the AI community, this work isolates the specific information capacity of raw gradients versus learned spatial filters, offering a rigorous reference point for evaluating feature representation efficiency in vision tasks.
A Sobel-Gradient MLP Baseline for Handwritten Character Recognition
arXiv:2508.11902v4 Announce Type: replace-cross Abstract: This study examines how much handwritten-character information is retained by a deliberately simple first-order edge representation. Instead of learning spatial filters, each input image is transformed by the fixed Sobel-Feldman operator into signed horizontal and vertical derivative maps, which are independently normalized, flattened, and classified by a multilayer perceptron (MLP). The resulting model therefore separates fixed edge extraction from learned classification and provides a controlled baseline for evaluating the sufficiency of first-order image gradients. In the executed experiments, the Sobel-gradient MLP achieves 98.54 percent test accuracy on MNIST and 92.50 percent on the TensorFlow Datasets (TFDS) EMNIST Letters configuration. Macro F1 scores are 0.9853 and 0.9265, respectively. One-vs-rest ROC analysis further yields micro/macro AUC values of 0.9998/0.9998 on MNIST and 0.9987/0.9982 on EMNIST Letters. Confusion-matrix analysis shows that the remaining errors are concentrated among geometrically similar classes, especially 3/8 and 4/9 for MNIST and I/L and G/Q for EMNIST Letters. These results show that fixed first-order gradients preserve substantial class-discriminative structure, while also revealing the specific ambiguities that remain when recognition is driven by edge geometry alone.
Abstract:This study examines how much handwritten-character information is retained by a deliberately simple first-order edge representation. Instead of learning spatial filters, each input image is transformed by the fixed Sobel-Feldman operator into signed horizontal and vertical derivative maps, which are independently normalized, flattened, and classified by a multilayer perceptron (MLP). The resulting model therefore separates fixed edge extraction from learned classification and provides a controlled baseline for evaluating the sufficiency of first-order image gradients. In the executed experiments, the Sobel-gradient MLP achieves 98.54 percent test accuracy on MNIST and 92.50 percent on the TensorFlow Datasets (TFDS) EMNIST Letters configuration. Macro F1 scores are 0.9853 and 0.9265, respectively. One-vs-rest ROC analysis further yields micro/macro AUC values of 0.9998/0.9998 on MNIST and 0.9987/0.9982 on EMNIST Letters. Confusion-matrix analysis shows that the remaining errors are concentrated among geometrically similar classes, especially 3/8 and 4/9 for MNIST and I/L and G/Q for EMNIST Letters. These results show that fixed first-order gradients preserve substantial class-discriminative structure, while also revealing the specific ambiguities that remain when recognition is driven by edge geometry alone.
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2508.11902 [cs.CV]
(or arXiv:2508.11902v4 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2508.11902 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 275 words
Related coverage
Robust Assamese Speech Recognition through Controlled Fine-Tuning of Whisper Models
arXiv · Jul 21, 2026
[AINews] AI Engineer NYC: Gemini 4 Argon, Vals AI Benchmarks
Latent Space · Oct 14, 2026
RefLAM: A Reference-Grounded Line Annotation Pipeline for Historical Arabic Manuscripts
arXiv · Aug 27, 2026
Handwritten Text Recognition Lives in the High-Pixel Variance Subspace
arXiv · Sep 30, 2026