2025-08 My first-author paper Superpixel-informed Continuous Low-Rank Tensor Representation for Multi-Dimensional Data Recovery have been under review at AAAI 2026!
Research (* indicates equal contribution, highlight indicates representative papers)
I'm interested in AI for Agentic LLMs, tensor representation and their applications.
Superpixel-informed Continuous Low-Rank Tensor Representation for Multi-Dimensional Data Recovery
A new framework, SCTR, which decomposes data into multiple semantically-consistent superpixels and learns a continuous, low-rank neural representation for each, significantly outperforming traditional methods in data recovery tasks.
Resources
Some tools and resources I've developed for the research community:
dir2prompts - A Python package for generating prompts from directory structure.