Our research spans multiple areas of artificial intelligence.
Explores techniques for updating or correcting specific knowledge in generative AI models without retraining them from scratch. Focuses on ensuring edits are accurate, reliable, localized, and aligned with safety and responsible AI principles.
Designing diffusion models and frameworks for medical image enhancement, disease progression modeling, synthetic chest X-ray generation, and non-invasive acoustic diagnosis.
Developing algorithms and video-centric datasets to improve object detection and image visibility under adverse weather conditions like fog and rain.
Utilizing AI for automatic question-answer generation, personalized teaching, student progress evaluation, and content extraction from scientific texts.
Focusing on digital media authenticity, invisible watermarking techniques, deepfake detection, and anti-spoofing frameworks.
Investigating and mitigating societal biases, cultural stereotypes, and demographic disparities in multimodal and large language models.