A dataset of deep learning performance from cross-base data encoding on MNIST and MNIST-C

Lawrence McKnight, Chandra Jaiswal, Issa AlHmoud, Balakrishna Gokaraju · 2024

Read A dataset of deep learning performance from cross-base data encoding on MNIST and MNIST-C by Lawrence McKnight, Chandra Jaiswal, Issa AlHmoud, Balakrishna Gokaraju (2024) 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/pmc-11697575.txt.

Description

Lawrence McKnight, Chandra Jaiswal, Issa AlHmoud, and Balakrishna Gokaraju publish a data set on how number bases affect a neural network's reading of handwritten digits, in a data paper from 2024. How data are represented can change how well a model learns. Novel representations are rarely studied in a systematic way.

The data set records a convolutional network's performance when the classic MNIST digit images, and a corrupted version of them, are re-encoded in many different numerical bases. Researchers can mine it for trends linking the base to accuracy. The authors hope it sparks wider tests of cross-base encoding on real-world tasks.

Catalog record

CREATORLawrence McKnight, Chandra Jaiswal, Issa AlHmoud, Balakrishna Gokaraju
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE26 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC11697575

Browse the full library · About the Tower of Babel