Four semifinalists for Regeneron Science Talent Search

“A Novel and Efficient Method of Persistent Homology to Detect and Remove Topological Errors in Triang Mesh Data” Modern algorithms take nearly 20 minutes to find and correct errors in 3D models of human brains generated from MRI and CT scans. Zhang improved upon these algorithms by more than six-fold by using patterns in 3D surface topology and geometry, shortening the runtime to only three m...

“An Investigation of the Formation of Fast radio Bursts (FRBs)” In his space-breaking research, Huang investigated the mysterious origins of FRBs, bursts of radio waves in the universe that have puzzled astrophysicists for nearly two decades. Huang’s research could revolutionize the way astrophysicists observe supernovae, calculate distance in the cosmos and study the origins of the unive...

“Single-cell Transcriptomic Analysis Reveals Genetic Drivers of of Fast/Slow Motor Neuron Identity” Motivated by his grandfather's paralysis due to amyotrophic lateral sclerosis, Luke Zhao discovered the genetic code for fast and slow motor neurons, which may one day provide answers to debilitating motor diseases, such as ALS and spinal muscular atrophy....

“Augmentation Strategies for Learning With Insufficient Data” In the data-driven world, algorithms require massive amounts of data in order to accurately model the world. However, many fields of research — rare diseases and uncommon objects — lack available data. Nishi combats this problem with transformations that magnify a small dataset by a thousand times....