Gagandeep Singh

Certifying neural networks ensures their safety and reliability in critical applications.

This article introduces a framework for analyzing program termination with improved efficiency.

A new training method enhances model safety against universal attacks.

New techniques enhance the verification of deep neural networks for safer applications.

A new method to evaluate bias in language models aims for fairer AI responses.

A system using MediaPipe and CNN to enhance sign language detection accuracy.

Discover how simple tweaks can trick chatbots into unexpected responses.

Automated methods improve neural network reliability through robust specifications.