Published in Peer Reviewed Core-A-ranking Computer Architecture Journal, 2024
This work introduces a lightweight adversarial detection filter that flags malicious inputs before they reach the model, offering an efficient and adaptable solution for enhancing the security of deep learning systems.
Download here
Published in Peer Reviewed Core-A-ranking Computer Architecture Journal, 2024
Transformers have served as a strong foundation in NLP and CV domain, but for practical application a dedicated hardware accelerator is required to increase throughput while keeping energy consumption lower than GPU.
Download here
Published in Proceedings of the AAAI Conference on Artificial Intelligence, 2023
GANs can now create realistic fake faces, but this research work explores how to track identity information leakage from real training data to the synthetic images.
Download here