Plote
Researcher • Scholar • Thinker
Academic Profile
Research Interests
My research interests lie in the theoretical foundations of generative modeling and their application to creating structurally coherent and physically plausible content. I approach this from a mathematical perspective, specifically studying diffusion and autoregressive models through their connections to Markov processes, stochastic differential equations, and Boltzmann-type distributions. My initial focus is on applying these principles to enhance structure-aware image generation, ensuring synthesized images are not just visually convincing but also compositionally sound. Building upon this, I extend these concepts from 2D to 3D, aiming to develop generative systems that understand the physical world. I am particularly intrigued by creating physically-grounded 3D models that possess an intrinsic understanding of their own geometry and material properties, which I see as a crucial step towards realistically modeling Human-Object Interaction (HOI). By ensuring models are grounded in physical reality, we can better simulate how objects respond to interaction. This same principle extends to resolving key challenges in video generation, where ensuring long-term temporal consistency remains a significant hurdle. My ultimate ambition is to contribute to the development of World Models by creating generative systems that unify rigorous mathematical modeling with a deep, intuitive grasp of physical dynamics and creative intelligence.
Biography
I am an undergraduate with a dual academic focus in Computer Science and a physics-intensive Materials Science program. My passion for physics lies in understanding complex systems by deriving their behavior from first principles. I particularly enjoy rigorously working through foundational derivations, from using statistical mechanics to derive the Maxwell-Boltzmann distribution and Planck’s law of black-body radiation, to exploring the fundamental postulates that lead to the Schrödinger equation in quantum mechanics. During my sophomore year, I delved into foundational computer science through hands-on systems projects. In compilers, I implemented a miniature front-end pipeline that transforms source code into an Abstract Syntax Tree (AST) and then into an intermediate representation, upon which I applied semantic analysis and basic optimizations. In operating systems, I extended the xv6 kernel through a series of experiments, engineering core functionalities such as copy-on-write fork to optimize memory usage, as well as sophisticated process schedulers and thread pool optimizations to ensure fair resource allocation under concurrent workloads. Over the past two years, my focus has converged on the intersection of machine learning, mathematical reasoning, and interactive 3D generation. I have explored the theoretical underpinnings of diffusion models by investigating concepts such as Schrödinger Bridges with my knowledge of mathematical statistics. More recently, my work has shifted towards application, where I am exploring 3D interactivity from a physics-based perspective. My current goal is to develop AI systems that can understand and interact with the world in a physically coherent manner.
Selected Publications
Affiliations
Academic Columns
笔记
Research notes and methodological insights from my ongoing projects and experiments.
论文介绍
Critical reviews and summaries of influential papers in AI and cognitive science.
关于我
Personal reflections on academic life, research philosophy, and interdisciplinary thinking.