A Sobel-Gradient MLP Baseline for Handwritten Character Recognition
Azam Nouri · 2025
Read A Sobel-Gradient MLP Baseline for Handwritten Character Recognition by Azam Nouri (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-11902.txt.
Description
Azam Nouri offers a simple baseline for reading handwritten characters in a 2025 paper. Rather than learning image filters, the study fixes them: a classic Sobel operator turns each image into horizontal and vertical edge maps. Those maps are scaled, flattened, and passed to a standard multilayer perceptron that does all the learning.
The setup scored 98.54 percent on the MNIST digit test set and 92.50 percent on a letters data set, with strong ranking scores on both. Remaining errors clustered among look-alike classes, such as three and eight among digits, and I and L or G and Q among letters. The results show how much class information plain first-order edges keep, and where edge geometry alone falls short.
Catalog record
| CREATOR | Azam Nouri |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 11 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-11902