Research programme

Projects & research.

Research studies, software and prototypes from across the NPerseus team.

Projects across the 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.

Thesis topics, funded studies and method work: attacks, defenses, data hiding, forensics and the datasets behind them.

32 matching research project records · Page 3 of 3

Member project Research

Multi-Signal Backdoor Defense research

Mahaveer Prasad

the published framework combines model signals to detect poisoned samples and correct their labels.

Link to this project
Member project Research

Protocol-Aware Robustness Evaluation research

Mahaveer Prasad

the published study evaluates deep-learning network intrusion-detection systems while preserving protocol semantics.

Link to this project
Member project Research

Quantum Computing and Cryptography Lab

Dr. Ankur

Developing the lab at NIT Delhi; the official laboratory-development record lists 2026.

Link to this project
Member project Research

Remote Sensing Cube Classification

Shray Gupta

applied contrastive learning to spectral-spatial feature extraction and hyperspectral cube classification.

Link to this project
Member project Research

Risk-Aware Poisoning Defense

Shray Gupta

developed a machine-learning framework for detecting and mitigating data-poisoning attacks.

Link to this project
Member project Research

ShortcutProbe research

Mahaveer Prasad

the published study investigates inference-time backdoor-sample detection through internal stability analysis.

Link to this project
Member project Research

Trust-Aware Underwater Routing

Ishan Panwar

contributed to a Q-learning-based routing system for underwater software-defined networks.

Link to this project
Member project Research

VPN-Detect Collector (vpn-collector-android)

Ayush Kumar

Built encrypted-traffic and VPN research software for collection coordination, traffic preparation, feature extraction, model training, evaluation and experiment provenance, using third-party tools and libraries. The work includes traffic contributed by Ayush Kumar and volunteers during everyday device use, kept distinct from imported benchmarks and synthetic examples. The study has documented consent and institutional ethics approval. Its public Android companion uses the PCAPdroid capture engine. A truncated public data edition is prepared locally and awaits publication; raw research access for trusted researchers and organisations is considered through enquiry and review.

Link to this projectPublic repository ↗
Browse all 74 profile project records
Explore the research areas that connect this work.

Security of Neural Networks

Defending deep models against adversarial examples, backdoor and data-poisoning attacks — and understanding when and why they fail.

Adversarial MLBackdoor DefenseRobustness

Data Hiding & Watermarking

Reversible data hiding in encrypted images, image and text steganography, and robust watermarking for ownership and integrity.

Reversible Data HidingSteganographyWatermarking

Multimedia Forensics

Detecting manipulated and GAN-generated media — deepfakes, splicing and median-filtering traces.

Deepfake DetectionImage ForensicsAuthenticity

Trustworthy & Private AI

Federated and privacy-preserving learning, explainable and trustworthy models, and methods for limiting the disclosure of training data.

Federated LearningExplainable AIPrivacy