Enhancing Deep Learning Model Robustness through Metamorphic Re-Training

Youssef Sameh Mostafa - Matriculation Number: 03788260, Karim Lotfy - Matriculation Number: 03788165, Aziz Said Togru - Matriculation Number: 12532570 (LMU) · 2024

Read Enhancing Deep Learning Model Robustness through Metamorphic Re-Training by Youssef Sameh Mostafa - Matriculation Number: 03788260, Karim Lotfy - Matriculation Number: 03788165, Aziz Said Togru - Matriculation Number: 12532570 (LMU) (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/arxiv-2412-01958.txt.

Description

Said Togru, Youssef Sameh Mostafa, and Karim Lotfy test whether rule-based retraining can make image models harder to fool, in a report from two thousand twenty-four. The method uses fixed rules about how a changed input should change the output, called metamorphic relations. Retraining runs in cycles with methods that learn from both labeled and unlabeled images.

Tests used three standard image sets and a mix of ready-made and fresh models. On average the score for toughness rose by about seventeen points for each model. Four named retraining methods were built into one shared frame so runs could repeat with ease. Readers get a hands-on route to tougher vision models for real use.

Catalog record

CREATORYoussef Sameh Mostafa - Matriculation Number: 03788260, Karim Lotfy - Matriculation Number: 03788165, Aziz Said Togru - Matriculation Number: 12532570 (LMU)
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE58 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-01958

Browse the full library · About the Tower of Babel