معرفی
Qiang Qiu is an Associate Professor in the Department of Electrical and Computer Engineering at Purdue University's College of Engineering. He actively contributes to the Vertically Integrated Projects (VIP) program, fostering interdisciplinary collaboration between faculty, students, and projects.
His research spans machine learning, computer vision, robotics, and generative AI, with particular emphasis on diffusion models, transformer fine-tuning, and graph neural networks. Key themes include improving model generalizability through orthogonal low-rank embeddings, developing efficient parameter-tuning strategies, and advancing visuotactile manipulation in robotics.
Recent publications reveal trends in attention control for text-to-image alignment, posterior sampling for diverse image generation, and federated learning security. His work addresses gradient conflicts in machine unlearning and explores multimodal sensing for robotic manipulation.
