Nicole Feng
Carnegie Mellon University
nfeng@andrew.cmu.edu
Bio
Nicole is a computer science PhD student at Carnegie Mellon University, where she develops algorithms for reliable computation with geometric data. Nicole received her B.S. in Applied and Computational Mathematics from the California Institute of Technology, and during her undergraduate studies she did research on fluid simulation and sketching. Nicole was named a WiGRAPH Rising Star in 2024, and in her spare time enjoys releasing graphics-related software and resources.
Areas of Research
- Computer Graphics and Vision
Robust Geometric Algorithms for Modern Computing
From natural phenomena (material structure, plant growth, geological formation, etc.) to human-generated data (manufactured objects, art, digital assets), our lives are defined by geometry. Algorithms for geometric problems enable modeling, simulation, and performance analysis, and thus are key to successfully manipulating the world around us: these algorithms drive technological progress across numerous domains including engineering, solid-modeling, animation, and other design-based industries. Yet we often lack complete answers to fundamental geometric questions such as, “How far is this point inside a given shape?” These queries are essential to tasks like reconstruction and robotic control, to name a few. On top of that, problems involving geometry are hard to solve, not because the problem itself is difficult to model or understand, but because of the difficulty in doing computation with geometry: even when underlying dynamics are well-understood, virtually all geometric data suffers quality issues, due to imperfect acquisition, reconstruction, and/or modeling. These low-level defects are unpredictable and easily frustrate higher-level design and optimization tasks such as reliable physics analysis or machine learning pipelines that demand perfect, clean data. Explicit repair of data is possible but time-consuming and tedious, with no guarantees of perfection; we instead need algorithms that allow direct computation of geometric quantities from defective data, which requires answering fundamental questions about geometry. My work identifies and answers these fundamental geometric questions, enabling reliable computation with geometric data. During my PhD, I’ve developed algorithms for robust inside/outside computation, signed distance, and animation tools.