Security of Neural Networks
Defending deep models against adversarial examples, backdoor and data-poisoning attacks — and understanding when and why they fail.
Research studies, software and prototypes from across the NPerseus team.
Research studies, prototypes and software described in public profiles. Status, roles and funding follow each source record; a profile entry does not by itself establish a released product.
74 matching profile project records · Page 4 of 7
DTU-recorded M.Tech (Research) thesis topic aligned with his two IEEE ICSCCC 2026 papers.
built an end-to-end regression pipeline with pandas, scikit-learn and gradient boosting, exposed through a FastAPI service.
Built experimental image-prediction and inpainting software, including training, checkpointing and evaluation tools, using third-party methods and libraries.
developed NodeMCU-based home automation and an LPG/smoke leakage detector.
Primary developer of Kartikeya, a Mindustry server-management plugin covering Discord integration, administration, voting, map workflows and player history.
co-announced a collaborative project focused on accessible conversational practice.
experimental Raspberry Pi/Linux driver project for light-based data communication, published as Laser-Driver.
A Data-Driven Game-Based Learning System for Ethical AI Use in Academia — UOW Dubai-funded project listed in 2025; AED 29,250 over 18 months. Pallavi Ranjan is a listed investigator alongside Farhad Oroumchian, Zeenath Reza Khan, Nkqubela Ruxwana, Dare Pitan, Veronika Krasnican and Mohamed Fawaz. Supports the Swiss MENA Project with CurveUp and the European Network for Academic Integrity. Current completion status is not stated.
a prototype retrieving PubMed literature and generating answers with retrieved sources.
developed a PHP and MySQL web portal for donating unused medicines.
Built experimental pipelines for model pruning and adapter-based recovery.
Defending deep models against adversarial examples, backdoor and data-poisoning attacks — and understanding when and why they fail.
Reversible data hiding in encrypted images, image and text steganography, and robust watermarking for ownership and integrity.
Detecting manipulated and GAN-generated media — deepfakes, splicing and median-filtering traces.
Federated and privacy-preserving learning, explainable and trustworthy models, and methods for limiting the disclosure of training data.