Over two millennia ago, Greek mathematician Euclid laid the groundwork for the modern understanding of geometry. Today, that work serves as the bedrock for researchers like Justin Solomon, who uses geometry to address complex problems – many of which seem unrelated to shapes at first glance. Solomon is an associate professor at MIT’s Department of Electrical Engineering and Computer Science (EECS) and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL).
A substantial part of Solomon’s work involves comparing high-dimensional datasets, to understand the impact on machine-learning models performance. Geometry plays an integral part in deciphering data arrangements and drawing useful comparisons, whether they involve distances, similarities, curvature, or shape. To that end, his Geometric Data Processing Group works on a diversity of issues, from aligning 3D organ scans for medical imaging to improving generative AI models.
Solomon’s interest in geometry and its applications was first sparked during his high school years, when he interned at a research lab developing algorithms for 3D face recognition. This experience led him to Stanford University, where he majored in maths and computer science, and landed a summer internship at Pixar Animation Studios. He worked with Pixar throughout his college years, focusing on physical simulation of cloth and fluids to enhance the realism of animated films, as well as rendering techniques.
After deciding to pursue academia, he stayed at Stanford for his PhD in computer science where he honed his attention on a problem known as ‘optimal transport’. Optimal transport focuses on shifting a distribution of items to another distribution as cheaply and efficiently as possible – much like trying to find the cost-effective way to transfer bags of flour from manufacturers to bakeries across a city.
Today, at MIT, Solomon continues to break down complex geometric problems. However, an equally important part of his role is reaching out to underrepresented students who typically have limited exposure to his field. He led the creation of the Summer Geometry Initiative – a six-week paid research program aimed at undergraduates, largely from underrepresented backgrounds. The initiative completed its third summer in 2023 and has been successful in changing the composition of incoming PhD students, not only at MIT but at other institutions as well.
In the future, Solomon hopes to apply geometric tools to enhance unsupervised machine learning models. These models must learn to recognize patterns without labelled training data, a challenge further complicated by the fact that most 3D data are not labelled. Nonetheless, by incorporating geometric insights, he believes it’s possible to help computers recognize patterns in complex, unlabeled 3D scenes more effectively.
When not solving geometric conundrums, Solomon indulges his love for classical music, playing the piano or cello. Both his musical practice and research are closely intertwined and mutually beneficial, with each discipline fueling his passion for the other.
